Low-precision training has emerged as a powerful technique for reducing computational and storage costs in Deep Neural Network (DNN) model training, enabling on-chip training or fine-tuning on edge devices. However, existing low-precision training methods often require higher bit-widths to maintain accuracy as model sizes increase. In this paper, we introduce an outlier-aware quantization strategy for low-precision training. While traditional value-aware quantization methods require costly online distribution statistics operations on computational data, impeding the efficiency gains of low-precision training, our approach addresses this challenge through a novel Learnable Threshold based Outlier-Aware Quantization (LT-OAQ) training framework. This method concurrently updates outlier thresholds and model weights through gradient descent, eliminating the need for costly data-statistics operations. To efficiently support the LT-OAQ training framework, we designed a hardware accelerator based on the systolic array architecture. This accelerator introduces a processing element (PE) fusion mechanism that dynamically fuses adjacent PEs into clusters to support outlier computations, optimizing the mapping of outlier computation tasks, enabling mixed-precision training, and implementing online quantization. Our approach maintains model accuracy while significantly reducing computational complexity and storage resource requirements. Experimental results demonstrate that our design achieves a 2.9 ×speedup in performance and a 2.17 ×reduction in energy consumption compared to state-of-the-art low-precision accelerators.

While a number of existing approaches for building foundation model agents rely on prompting or fine-tuning with human demonstrations, it is not sufficient in dynamic environments (e.g., mobile device control). On-policy reinforcement learning (RL) should address these limitations, but collecting actual rollouts in an environment is often undesirable in truly open-ended agentic problems such as mobile device control or interacting with humans, where each unit of interaction is associated with a cost. In such scenarios, a method for policy learning that can utilize off-policy experience by learning a trained action-value function is much more effective. In this paper, we develop an approach, called Digi-Q, to train VLM-based action-value Q-functions which are then used to extract the agent policy. We study our approach in the mobile device control setting. Digi-Q trains the Q-function using offline temporal-difference (TD) learning, on top of frozen, intermediate-layer features of a VLM. Compared to fine-tuning the whole VLM, this approach saves us compute and enhances scalability. To make the VLM features amenable for representing the Q-function, we need to employ an initial phase of fine-tuning to amplify coverage over actionable information needed for value function. Once trained, we use this Q-function via a Best-of-N policy extraction operator that imitates the best action out of multiple candidate actions from the current policy as ranked by the value function, enabling policy improvement without environment interaction. Digi-Q outperforms several prior methods on user-scale device control tasks in Android-in-the-Wild, attaining 21.2% improvement over prior best-performing method. In some cases, our Digi-Q ap- proach already matches state-of-the-art RL methods that require interaction. The project is open-sourced at https://github.com/DigiRL-agent/digiq

A generalist foundation model agent needs to have a large and diverse skill repertoire, such as finding directions between two travel locations and buying specific items from the Internet. If each skill needs to be specified manually through a fixed set of human-annotated instructions, the agent’s skill repertoire will necessarily be limited due to the scalability of human-annotated instructions. In this work, we address this challenge by proposing Proposer-Agent-Evaluator (PAE), an effective learning system that enables foundation model agents to autonomously discover and practice skills in the wild. After a context-aware task proposer generates instructions based on website information, the agent policy attempts those tasks in the real world with resulting trajectories evaluated by an autonomous VLM-based success evaluator. The success evaluation serves as the reward signal for the agent to refine its policies through RL. We validate PAE on challenging vision-based web navigation, using both real-world and selfhosted websites from WebVoyager and WebArena. Our results show that PAE significantly improves the zero-shot generalization capability of VLM Internet agents (around 50% relative improvement) to both unseen tasks and websites.

Tianyang Zhao, Kunwar Yashraj Singh, Srikar Appalaraju, Peng Tang, Ying Nian Wu, Li Erran Li. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Diffusion Transformers (DiTs) have recently demonstrated remarkable performance in visual generation tasks, surpassing traditional U-Net-based diffusion models by significantly improving image and video generation quality and scalability. However, the large model size and iterative denoising process introduce substantial computational and memory overhead, limiting their deployment in real-world applications. Post-training quantization (PTQ) is a promising solution that compresses models and accelerates inference by converting weights and activations to low-bit representations. Despite its potential, PTQ faces significant challenges when applied to DiTs, often resulting in severe degradation of generative quality. To address these issues, we propose VETA-DiT (**V**ariance-**E**qualized and **T**emporal **A**daptation for **Di**ffusion **T**ransformers), a dedicated quantization framework for DiTs. Our method first analyzes the sources of quantization error from the perspective of inter-channel variance and introduces a Karhunen–Loève Transform enhanced alignment to equalize variance across channels, facilitating effective quantization under low bit-widths. Furthermore, to handle the temporal variation of activation distributions inherent in the iterative denoising steps of DiTs, we design an incoherence-aware adaptive method that identifies and properly calibrates timesteps with high quantization difficulty. We validate VETA-DiT on extensive image and video generation tasks, preserving acceptable visual quality under the more aggressive W4A4 configuration. Specifically, VETA-DiT reduces FID by 33.65 on the DiT-XL/2 model and by 45.76 on the PixArt-$\Sigma$ model compared to the baseline under W4A4, demonstrating its strong quantization capability and generative performance. Code is available at: https://github.com/xululi0223/VETA-DiT.

Abstract Despite the recent impressive breakthroughs in text-to-image generation, generative models have difficulty in capturing the data distribution of underrepresented attribute compositions while over-memorizing overrepresented attribute compositions, which raises public concerns about their robustness and fairness. To tackle this challenge, we propose ACTIG, an attribute-centric compositional text-to-image generation framework. We present an attribute-centric feature augmentation and a novel image-free training scheme, which greatly improves model’s ability to generate images with underrepresented attributes. We further propose an attribute-centric contrastive loss to avoid overfitting to overrepresented attribute compositions. We validate our framework on the CelebA-HQ and CUB datasets. Extensive experiments show that the compositional generalization of ACTIG is outstanding, and our framework outperforms previous works in terms of image quality and text-image consistency. The source code and trained models are publicly available at https://github.com/yrcong/ACTIG .

Most pre-trained learning systems are known to suffer from bias, which typically emerges from the data, the model, or both. Measuring and quantifying bias and its sources is a challenging task and has been extensively studied in image captioning. Despite the significant effort in this direction, we observed that existing metrics lack consistency in the inclusion of the visual signal. In this paper, we introduce a new bias assessment metric, dubbed ImageCaptioner2, for image captioning. Instead of measuring the absolute bias in the model or the data, ImageCaptioner2pay more attention to the bias introduced by the model w.r.t the data bias, termed bias amplification. Unlike the existing methods, which only evaluate the image captioning algorithms based on the generated captions only, ImageCaptioner2incorporates the image while measuring the bias. In addition, we design a formulation for measuring the bias of generated captions as prompt-based image captioning instead of using language classifiers. Finally, we apply our ImageCaptioner2metric across 11 different image captioning architectures on three different datasets, i.e., MS-COCO caption dataset, Artemis V1, and Artemis V2, and on three different protected attributes, i.e., gender, race, and emotions. Consequently, we verify the effectiveness of our ImageCaptioner2metric by proposing Anonymous-Bench, which is a novel human evaluation paradigm for bias metrics. Our metric shows significant superiority over the recent bias metric; LIC, in terms of human alignment, where the correlation scores are 80% and 54% for our metric and LIC, respectively. The code and more details are available at https://eslambakr.github.io/imagecaptioner2.github.io/.

Learning commonsense reasoning from visual contexts and scenes in real-world is a crucial step toward advanced artificial intelligence. However, existing video reasoning benchmarks are still inadequate since they were mainly designed for factual or situated reasoning and rarely involve broader knowledge in the real world. Our work aims to delve deeper into reasoning evaluations, specifically within dynamic, open-world, and structured context knowledge. We propose a new benchmark (SOK-Bench), consisting of 44K questions and 10K situations with instance-level annotations depicted in the videos. The reasoning process is required to understand and apply situated knowledge and general knowledge for problem-solving. To create such a dataset, we propose an automatic and scalable gener-ation method to generate question-answer pairs, knowledge graphs, and rationales by instructing the combinations of LLMs and MLLMs. Concretely, we first extract observable situated entities, relations, and processes from videos for situated knowledge and then extend to open-world knowledge beyond the visible content. The task generation is facilitated through multiple dialogues as iterations and subsequently corrected and refined by our designed self-promptings and demonstrations. With a corpus of both explicit situated facts and implicit commonsense, we generate associated question-answer pairs and reasoning processes, finally followed by manual reviews for quality assurance. We evaluated recent mainstream large vision-language models on the benchmark and found several in-sightful conclusions. For more information, please refer to our benchmark at www.bobbywu.com/SOKBench.

Object detection tasks are central to the development of datasets and algorithms in computer vision and machine learning. Despite its centrality, object detection remains tedious and time-consuming due to the inherent interactions that are often associated with drawing precise annotations. In this paper, we introduce Snapper, an interactive and intelligent annotation tool that intercepts bounding box annotations as they’re drawn and “snaps” them to the nearby object edges in real-time. Through a mixed-design user study with 18 full-time annotators, we compare Snapper’s annotation mode to alternative modes of annotation and find that Snapper enables participants to complete object detection tasks 39% more quickly without diminishing annotation quality. Further, we find that participants perceive Snapper as a tool that is interactively intuitive, trustworthy, and helpful. We conclude by discussing the implications of our findings as they relate to augmenting annotators’ conventions for drawing annotations in practice.

We aim to evaluate Large Language Models (LLMs) for embodied decision making. While a significant body of work has been leveraging LLMs for decision making in embodied environments, we still lack a systematic understanding of their performance because they are usually applied in different domains, for different purposes, and built based on different inputs and outputs. Furthermore, existing evaluations tend to rely solely on a final success rate, making it difficult to pinpoint what ability is missing in LLMs and where the problem lies, which in turn blocks embodied agents from leveraging LLMs effectively and selectively. To address these limitations, we propose a generalized interface (Embodied Agent Interface) that supports the formalization of various types of tasks and input-output specifications of LLM-based modules. Specifically, it allows us to unify 1) a broad set of embodied decision-making tasks involving both state and temporally extended goals, 2) four commonly-used LLM-based modules for decision making: goal interpretation, subgoal decomposition, action sequencing, and transition modeling, and 3) a collection of fine-grained metrics that break down evaluation into error types, such as hallucination errors, affordance errors, and various types of planning errors. Overall, our benchmark offers a comprehensive assessment of LLMs’ performance for different subtasks, pinpointing the strengths and weaknesses in LLM-powered embodied AI systems and providing insights into the effective and selective use of LLMs in embodied decision making.

Recent progress on vision-language foundation models have brought significant advancement to building generalpurpose robots. By using the pre-trained models to encode the scene and instructions as inputs for decision making, the instruction-conditioned policy can generalize across different objects and tasks. While this is encouraging, the policy still fails in most cases given an unseen task or environment. In this work, we propose Policy Adaptation from Foundation model Feedback (PAFF). When deploying the trained policy to a new task or a new environment, we first let the policy play with randomly generated instructions to record the demonstrations. While the execution could be wrong, we can use the pre-trained foundation models to provide feedback to relabel the demonstrations. This automatically provides new pairs of demonstration-instruction data for policy fine-tuning. We evaluate our method on a broad range of experiments with the focus on generalization on unseen objects, unseen tasks, unseen environments, and sim-to-real transfer. We show PAFF improves baselines by a large margin in all cases.

Self-supervised pretraining on large unlabeled datasets has shown tremendous success in improving the task performance of many 2D and small scale 3D computer vision tasks. However, the popular pretraining approaches have not been impactfully applied to outdoor LiDAR point cloud perception due to the latter's scene complexity and wide range. We propose a new self-supervised pretraining method ISCC with two novel pretext tasks for LiDAR point clouds. The first task uncovers semantic information by sorting local groups of points in the scene into a globally consistent set of semantically meaningful clusters using contrastive learning, complemented by a second task which reasons about precise surfaces of various parts of the scene through implicit surface reconstruction to learn geometric structures. We demonstrate their effectiveness through transfer learning on 3D object detection and semantic segmentation in real world LiDAR scenes. We further design an unsupervised semantic grouping task to show that our approach learns highly semantically meaningful features.

In recent years, Text-to-Image (T2I) models have been extensively studied, especially with the emergence of diffusion models that achieve state-of-the-art results on T2I synthesis tasks. However, existing benchmarks heavily rely on subjective human evaluation, limiting their ability to holistically assess the model’s capabilities. Furthermore, there is a significant gap between efforts in developing new T2I architectures and those in evaluation. To address this, we introduce HRS-Bench, a concrete evaluation benchmark for T2I models that is Holistic, Reliable, and Scalable. Unlike existing benchmarks that focus on limited aspects, HRS-Bench measures 13 skills that can be categorized into five major categories: accuracy, robustness, generalization, fairness, and bias. In addition, HRS-Bench covers 50 scenarios, including fashion, animals, transportation, food, and clothes. We evaluate nine recent large-scale T2I models using metrics that cover a wide range of skills. A human evaluation aligned with 95% of our evaluations on average was conducted to probe the effectiveness of HRS-Bench. Our experiments demonstrate that existing models often struggle to generate images with the desired count of objects, visual text, or grounded emotions. We hope that our benchmark help ease future text-to-image generation research. The code and data are available at https://eslambakr.github.io/hrsbench.github.io/.

In recent years, increasing attention has been directed to leveraging pre-trained vision models for motor control. While existing works mainly emphasize the importance of this pre-training phase, the arguably equally important role played by downstream policy learning during control-specific fine-tuning is often neglected. It thus remains unclear if pre-trained vision models are consistent in their effectiveness under different control policies. To bridge this gap in understanding, we conduct a comprehensive study on 14 pre-trained vision models using 3 distinct classes of policy learning methods, including reinforcement learning (RL), imitation learning through behavior cloning (BC), and imitation learning with a visual reward function (VRF). Our study yields a series of intriguing results, including the discovery that the effectiveness of pre-training is highly dependent on the choice of the downstream policy learning algorithm. We show that conventionally accepted evaluation based on RL methods is highly variable and therefore unreliable, and further advocate for using more robust methods like VRF and BC. To facilitate more universal evaluations of pre-trained models and their policy learning methods in the future, we also release a benchmark of 21 tasks across 3 different environments alongside our work.

Transformers have recently shown superior performances on various vision tasks. The large, sometimes even global, receptive field endows Transformer models with higher representation power over their CNN counterparts. Nevertheless, simply enlarging receptive field also gives rise to several concerns. On the one hand, using dense attention e.g., in ViT, leads to excessive memory and computational cost, and features can be influenced by irrelevant parts which are beyond the region of interests. On the other hand, the sparse attention adopted in PVT or Swin Transformer is data agnostic and may limit the ability to model long range relations. To mitigate these issues, we propose a novel deformable selfattention module, where the positions of key and value pairs in selfattention are selected in a data-dependent way. This flexible scheme enables the self-attention module to focus on relevant re-gions and capture more informative features. On this basis, we present Deformable Attention Transformer, a general backbone model with deformable attention for both image classification and dense prediction tasks. Extensive experi-ments show that our models achieve consistently improved results on comprehensive benchmarks. Code is available at https://github.com/LeapLabTHU/DAT.

Recent work has seen the development of general purpose neural architectures that can be trained to perform tasks across diverse data modalities. General purpose models typically make few assumptions about the underlying data-structure and are known to perform well in the large-data regime. At the same time, there has been growing interest in modular neural architectures that represent the data using sparsely interacting modules. These models can be more robust out-of-distribution, computationally efficient, and capable of sample-efficient adaptation to new data. However, they tend to make domain-specific assumptions about the data, and present challenges in how module behavior (i.e., parameterization) and connectivity (i.e., their layout) can be jointly learned. In this work, we introduce a general purpose, yet modular neural architecture called Neural Attentive Circuits (NACs) that jointly learns the parameterization and a sparse connectivity of neural modules without using domain knowledge. NACs are best understood as the combination of two systems that are jointly trained end-to-end: one that determines the module configuration and the other that executes it on an input. We demonstrate qualitatively that NACs learn diverse and meaningful module configurations on the Natural Language and Visual Reasoning for Real (NLVR2) dataset without additional supervision. Quantitatively, we show that by incorporating modularity in this way, NACs improve upon a strong non-modular baseline in terms of low-shot adaptation on CIFAR and Caltech-UCSD Birds dataset (CUB) by about 10 percent, and OOD robustness on Tiny ImageNet-R by about 2.5 percent. Further, we find that NACs can achieve an 8x speedup at inference time while losing less than 3 percent performance. Finally, we find NACs to yield competitive results on diverse data modalities spanning point-cloud classification, symbolic processing and text-classification from ASCII bytes, thereby confirming its general purpose nature.

Pretraining on large unlabeled datasets has been proven to improve the down-stream task performance on many computer vision tasks, such as 2D object detection and video classification. However, for large-scale 3D scenes, such as outdoor LiDAR point clouds, pretraining is not widely used. Due to the special data characteristics of large 3D point clouds, 2D pretraining frameworks tend to not generalize well. In this paper, we propose a new self-supervised pretraining method that targets large-scale 3D scenes. We pretrain commonly used point-based and voxel-based model architectures and show the transfer learning performance on 3D object detection and also semantic segmentation. We demonstrate the effectiveness of our approach on both dense 3D indoor point clouds and also sparse outdoor lidar point clouds.

Feature learning for 3D object detection from point clouds is very challenging due to the irregularity of 3D point cloud data. In this paper, we propose Pointformer, a Transformer backbone designed for 3D point clouds to learn features effectively. Specifically, a Local Transformer module is employed to model interactions among points in a local region, which learns context-dependent region features at an object level. A Global Transformer is designed to learn context-aware representations at the scene level. To further capture the dependencies among multi-scale representations, we propose Local-Global Transformer to integrate local features with global features from higher resolution. In addition, we introduce an efficient coordinate refinement module to shift down-sampled points closer to object centroids, which improves object proposal generation. We use Pointformer as the backbone for state-of-the-art object detection models and demonstrate significant improvements over original models on both indoor and outdoor datasets.

Vehicle detection with visual sensors like lidar and camera is one of the critical functions enabling autonomous driving. While they generate fine-grained point clouds or high-resolution images with rich information in good weather conditions, they fail in adverse weather (e.g., fog) where opaque particles distort lights and significantly reduce visibility. Thus, existing methods relying on lidar or camera experience significant performance degradation in rare but critical adverse weather conditions. To remedy this, we resort to exploiting complementary radar, which is less impacted by adverse weather and becomes prevalent on vehicles. In this paper, we present Multimodal Vehicle Detection Network (MVDNet), a two-stage deep fusion detector, which first generates proposals from two sensors and then fuses region-wise features between multimodal sensor streams to improve final detection results. To evaluate MVDNet, we create a procedurally generated training dataset based on the collected raw lidar and radar signals from the open-source Oxford Radar Robotcar. We show that the proposed MVDNet surpasses other state-of-the-art methods, notably in terms of Average Precision (AP), especially in adverse weather conditions. The code and data are available at https://github.com/qiank10/MVDNet.

Monocular 3D prediction is one of the fundamental problems in 3D vision. Recent deep learning-based approaches have brought us exciting progress on this problem. However, existing approaches have predominantly focused on end-to-end depth and normal predictions, which do not fully utilize the underlying 3D environment’s geometric structures. This paper introduces StruMonoNet, which detects and enforces a planar structure to enhance pixel-wise predictions. StruMonoNet innovates in leveraging a hybrid representation that combines visual feature and a surfel representation for plane prediction. This formulation allows us to combine the power of visual feature learning and the flexibility of geometric representations in incorporating geometric relations. As a result, StruMonoNet can detect relations between planes such as adjacent planes, perpendicular planes, and parallel planes, all of which are beneficial for dense 3D prediction. Experimental results show that StruMonoNet considerably outperforms state-of-the-art approaches on NYUv2 and ScanNet.

Spoken language understanding (SLU) is the task of inferring the semantics of spoken utterances. Traditionally, this has been achieved with a cascading combination of Automatic Speech Recognition (ASR) and Natural Language Understanding (NLU) modules that are optimized separately, which can lead to a suboptimal overall performance. More recently, End-to-End SLU (E2E SLU) was proposed to perform SLU directly from speech through a joint optimization of the modules, addressing some of the traditional SLU shortcomings. A key challenge of this approach is how to best integrate the feature learning of the ASR and NLU sub-tasks to maximize their performance. While it is known that in general, ASR models focus on low-level features, and NLU models need higher-level contextual information, ASR models can nonetheless also leverage top-down syntactic and semantic information to improve their recognition. Based on this insight, we propose Top-Down SLU (TD-SLU), a new transformer-based E2E SLU model that uses top-down attention and an attention gate to fuse high-level NLU features with low-level ASR features, which leads to a better optimization of both tasks. We have validated our model using the public FluentSpeech set, and a large custom dataset. Results show TD-SLU is able to outperform selected baselines both in terms of ASR and NLU quality metrics, and suggest that the added syntactic and semantic high-level information can improve the model’s performance.

Motion prediction of vehicles is critical but challenging due to the uncertainties in complex environments and the limited visibility caused by occlusions and limited sensor ranges. In this paper, we study a new task, safety-aware motion prediction with unseen vehicles for autonomous driving. Unlike the existing trajectory prediction task for seen vehicles, we aim at predicting an occupancy map that indicates the earliest time when each location can be occupied by either seen and unseen vehicles. The ability to predict unseen vehicles is critical for safety in autonomous driving. To tackle this challenging task, we propose a safety-aware deep learning model with three new loss functions to predict the earliest occupancy map. Experiments on the large-scale autonomous driving nuScenes dataset show that our proposed model significantly outperforms the state-of-the-art baselines on the safety-aware motion prediction task. To the best of our knowledge, our approach is the first one that can predict the existence of unseen vehicles in most cases. Project page at https://github.com/xrenaa/Safety-Aware-Motion-Prediction.

Masked language models have revolutionized natural language processing systems in the past few years. A recently introduced generalization of masked language models called warped language models are trained to be more robust to the types of errors that appear in automatic or manual transcriptions of spoken language by exposing the language model to the same types of errors during training. In this work we propose a novel approach that takes advantage of the robustness of warped language models to transcription noise for correcting transcriptions of spoken language. We show that our proposed approach is able to achieve up to 10% reduction in word error rates of both automatic and manual transcriptions of spoken language.

Controllable text generation concerns two fundamental tasks of wide applications, namely generating text of given attributes (i.e., attribute-conditional generation), and minimally editing existing text to possess desired attributes (i.e., text attribute transfer). Extensive prior work has largely studied the two problems separately, and developed different conditional models which, however, are prone to producing biased text (e.g., various gender stereotypes). This paper proposes to formulate controllable text generation from a principled causal perspective which models the two tasks with a unified framework. A direct advantage of the causal formulation is the use of rich causality tools to mitigate generation biases and improve control. We treat the two tasks as interventional and counterfactual causal inference based on a structural causal model, respectively. We then apply the framework to the challenging practical setting where confounding factors (that induce spurious correlations) are observable only on a small fraction of data. Experiments show significant superiority of the causal approach over previous conditional models for improved control accuracy and reduced bias.

We present Uncertainty-aware Cascaded Stereo Network (UCS-Net) for 3D reconstruction from multiple RGB images. Multi-view stereo (MVS) aims to reconstruct fine-grained scene geometry from multi-view images. Previous learning-based MVS methods estimate per-view depth using plane sweep volumes (PSVs) with a fixed depth hypothesis at each plane; this requires densely sampled planes for high accuracy, which is impractical for high-resolution depth because of limited memory. In contrast, we propose adaptive thin volumes (ATVs); in an ATV, the depth hypothesis of each plane is spatially varying, which adapts to the uncertainties of previous per-pixel depth predictions. Our UCS-Net has three stages: the first stage processes a small PSV to predict low-resolution depth; two ATVs are then used in the following stages to refine the depth with higher resolution and higher accuracy. Our ATV consists of only a small number of planes with low memory and computation costs; yet, it efficiently partitions local depth ranges within learned small uncertainty intervals. We propose to use variance-based uncertainty estimates to adaptively construct ATVs; this differentiable process leads to reasonable and fine-grained spatial partitioning. Our multi-stage framework progressively sub-divides the vast scene space with increasing depth resolution and precision, which enables reconstruction with high completeness and accuracy in a coarse-to-fine fashion. We demonstrate that our method achieves superior performance compared with other learning-based MVS methods on various challenging datasets.

In the domain of autonomous driving, deep learning has substantially improved the 3D object detection accuracy for LiDAR and stereo camera data alike. While deep networks are great at generalization, they are also notorious to overfit to all kinds of spurious artifacts, such as brightness, car sizes and models, that may appear consistently throughout the data. In fact, most datasets for autonomous driving are collected within a narrow subset of cities within one country, typically under similar weather conditions. In this paper we consider the task of adapting 3D object detectors from one dataset to another. We observe that naively, this appears to be a very challenging task, resulting in drastic drops in accuracy levels. We provide extensive experiments to investigate the true adaptation challenges and arrive at a surprising conclusion: the primary adaptation hurdle to overcome are differences in car sizes across geographic areas. A simple correction based on the average car size yields a strong correction of the adaptation gap. Our proposed method is simple and easily incorporated into most 3D object detection frameworks. It provides a first baseline for 3D object detection adaptation across countries, and gives hope that the underlying problem may be more within grasp than one may have hoped to believe. Our code is available at https://github. com/cxy1997/3D_adapt_auto_driving.

Recently, autonomous driving ignited competitions among car makers and technical corporations. Low-level autonomous vehicles are already commercially available. However, high autonomous vehicles where the vehicle drives by itself without human monitoring is still at infancy. Such autonomous vehicles (AVs) fully rely on the computing system in the car to perceive the environment and make driving decisions. In AV computing systems, the latency is an essential metric for ensuring the efficiency and safety, because a timely decision with low latency will avoid accidents and save lives. Moreover, we perform a field study by running industrial Level-4 autonomous driving fleets in various locations, road conditions, and traffic patterns. We observe that the perception module consumes the longest latency, and it is highly sensitive to surrounding obstacles. To study the correlation between perception latency and surrounding obstacles, we propose a perception latency model. Moreover, we demonstrate the use of our latency model, by developing and evaluating a driving scenario perception-aware AV computing system that efficiently manages computation hardware resource. Our evaluation results show that the proposed AV system resource management improves performance significantly.

Depth from a monocular video can enable billions of devices and robots with a single camera to see the world in 3D. In this paper, we present a model for video depth estimation, which consists of a flow-to-depth layer, a camera pose refinement module, and a depth fusion network. Given optical flow and camera poses, our flow-to-depth layer generates depth proposals and their corresponding confidence maps by explicitly solving an epipolar geometry optimization problem. Our flow-to-depth layer is differentiable, and thus we can refine camera poses by maximizing the aggregated confidence in the camera pose refinement module. Our depth fusion network can utilize the target frame, depth proposals, and confidence maps inferred from different neighboring frames to produce the final depth map. Furthermore, the depth fusion network can additionally take the depth proposals generated by other methods to further improve the results. The experiments on three public datasets show that our approach outperforms state-of-the-art depth estimation methods, and has reasonable crossdataset generalization ability: our model trained on KITTI still performs well on the unseen Waymo dataset.

High automated vehicles rely on the computing system in the car to understand the environment and make driving decisions. Therefore, computing system design is essential for ensuring the driving safety. However, to our knowledge, no clear guideline exists so far regarding how to guide the safety-aware autonomous vehicle (AV) computing system design. To understand the safety requirement of AV computing system, we performed a field study by operating industrial Level-4 AV fleets in multiple locations for three months. The field study indicates that traditional computing system performance metrics, such as tail latency, average latency, maximum latency, and timeout, cannot fully satisfy the safety requirement for AV computing system design. To address this issue, we propose the “safety score” as a primary metric for measuring the level of safety in AV computing system design.

An increasing amount of mobile analytics is performed on data that is procured in a real-time fashion to make real-time decisions. Such tasks include simple reporting on streams to sophisticated model building. However, the practicality of these analyses are impeded in several domains because they are faced with a fundamental trade-off between data collection latency and analysis accuracy. In this paper, we first study this trade-off in the context of a specific domain, Cellular Radio Access Networks (RAN). We find that the trade-off can be resolved using two broad, general techniques: intelligent data grouping and task formulations that leverage domain characteristics. Based on this, we present CellScope, a system that applies a domain specific formulation and application of Multi-task Learning (MTL) to RAN performance analysis. It uses three techniques: feature engineering to transform raw data into effective features, a PCA inspired similarity metric to group data from geographically nearby base stations sharing performance commonalities, and a hybrid online-offline model for efficient model updates. Our evaluation shows that CellScope's accuracy improvements over direct application of ML range from 2.5× to 4.4× while reducing the model update overhead by up to 4.8×. We have also used CellScope to analyze an LTE network of over 2 million subscribers, where it reduced troubleshooting efforts by several magnitudes. We then apply the underlying techniques in CellScope to another domain specific problem, mobile phone energy bug diagnosis, and show that the techniques are general.

In an increasingly mobile connected world, our user experience of mobile applications more and more depends on the performance of cellular radio access networks (RAN). To achieve high quality of experience for the user, it is imperative that operators identify and diagnose performance problems quickly. In this paper, we describe our experience in understanding the challenges in automating the diagnosis of RAN performance problems. Working with a major cellular network operator on a part of their RAN that services more than 2 million users, we demonstrate that fine-grained modeling and analysis could be the key towards this goal. We describe our methodology in analyzing RAN problems, and highlight a few of our findings, some previously unknown. We also discuss lessons from our attempt at building automated diagnosis solutions.

Software-defined networking (SDN) promises unprecedentedly flexible network management but it is susceptible to forwarding faults. Such faults originate from data-plane rules with missing faults and priority faults. Yet existing fault detection ignores priority faults, because they are not discovered on commercial switches until recently. In this paper, we present RuleScope, a more comprehensive solution for inspecting SDN forwarding. RuleScope offers a series of accurate and efficient algorithms for detecting and troubleshooting rule faults. They inspect forwarding behavior using customized probe packets to exercise data-plane rules. The detection algorithm exposes not only missing faults but also priority faults and the troubleshooting algorithm uncover actual forwarding states of data-plane flow tables. Both of them help track real-time forwarding status and benefit reliable network monitoring. Furthermore, toward fast inspection of dynamic networks, we propose incremental algorithms for rapidly evolving network policies to amortize detection and troubleshooting overhead without sacrificing accuracy. Experiments with our prototype on the Ryu SDN controller and Pica8 P-3297 switch show that the RuleScope achieves accurate fault detection on 320-entry flow tables with a cost of 1500+ probe packets within 16 s.

Software-dehned network (SDN) is deemed to enable more dynamic management of data center networks that promptly respond to network events with changes in network policies. Although the SDN controller architecture is increasingly optimized for swift policy updates, the data plane, especially the prevailing TCAM-based flow tables on physical SDN switches, remains unoptimized for fast rule updates, and is gradually becoming the primary bottleneck along the policy update pipeline. In this paper, we present RuleTris, the hrst SDN update optimization framework that minimizes rule update latency for TCAM-based switches. RuleTris employs the dependency graph (DAG) as the key abstraction to minimize the update latency. RuleTris efhciently obtains the DAGs with novel dependency preserving algorithms that incrementally build rule dependency along with the compilation process. Then, in the guidance of the DAG, RuleTris optimizes the rule updates in TCAM to avoid unnecessary entry moves, which are the main cause of TCAM update inefhciency. We prove that RuleTris generates TCAM updates with the minimum number of TCAM entry moves. In evaluation, RuleTris achieves a median of <;12ms and 90-percentile of <;15ms the end-to-end per-rule update latency on our hardware prototype, outperforming the state-of-the-art composition compiler CoVisor by ~20 times.

Software-Defined Networking (SDN) promises un-precedentedly flexible network management but it is susceptible to forwarding faults. Such faults originate from data-plane rules with missing faults and priority faults. Yet existing fault detection ignores priority faults because they are not discovered on commercial switches until recently. In this paper, we present RuleScope, a more comprehensive solution for inspecting SDN forwarding. RuleScope offers a series of accurate and efficient algorithms for detecting and troubleshooting rule faults. They inspect forwarding behavior using customized probe packets to exercise data-plane rules. The detection algorithm exposes not only missing faults but also priority faults. Beyond simply detecting rule faults, the troubleshooting algorithms uncover actual data-plane flow tables. They help track real-time forwarding status and benefit reliable network monitoring. We explore various techniques for enhancing algorithm efficiency without sacrificing inspection accuracy. Experiments with our prototype on the Ryu SDN controller and Pica8 P-3297 switch show that RuleScope achieves accurate and efficient forwarding inspection with limited bandwidth and packet-switching overhead.

Adaptive HTTP video streaming over LTE has been gaining popularity due to LTE's high capacity. Quality of adaptive streaming depends highly on the accuracy of client's estimation of end-to-end network bandwidth, which is challenging due to LTE link dynamics. In this paper, we present piStream, that allows a client to efficiently monitor the LTE basestation's PHY-layer resource allocation, and then map such information to an estimation of available bandwidth. Given the PHY-informed bandwidth estimation, piStream uses a probabilistic algorithm to balance video quality and the risk of stalling, taking into account the burstiness of LTE downlink traffic loads. We conduct a real-time implementation of piStream on a software-radio tethered to an LTE smartphone. Comparison with state-of-the-art adaptive streaming protocols demonstrates that piStream can effectively utilize the LTE bandwidth, achieving high video quality with minimal stalling rate.

We conduct a comprehensive measurement study of switch control plane latencies using four types of production SDN switches. Our measurements show that control actions, such as rule installation, have surprisingly high latency, due to both software implementation inefficiencies and fundamental traits of switch hardware. We also propose three measurement-driven latency mitigation techniques---optimizing route selection, spreading rules across switches, and reordering rule installations---to effectively tame the flow setup latencies in SDN.

A major benefit of software-defined networking (SDN) over traditional networking is simpler and easier control of network devices. The diversity of SDN switch implementation properties, which include both diverse switch hardware capabilities and diverse control-plane software behaviors, however, can make it difficult to understand and/or to control the switches in an SDN network. In this paper, we present Tango, a novel framework to explore the issues of understanding and optimization of SDN control, in the presence of switch diversity. The basic idea of Tango is novel, simple, and yet quite powerful. In particular, different from all previous SDN control systems, which either ignore switch diversity or depend on that switches can and will report diverse switch implementation properties, Tango introduces a novel, proactive probing engine that infers key switch capabilities and behaviors, according to a well-structured set of Tango patterns, where a Tango pattern consists of a sequence of standard OpenFlow commands and a corresponding data traffic pattern. Utilizing the inference results from Tango patterns and additional application API hints, Tango conducts automatic switch control optimization, despite diverse switch capabilities and behaviors. Evaluating Tango on both hardware switches and emulated software switches, we show that Tango can infer flow table sizes, which are key switch implementation properties, within less than 5% of actual values, despite diverse switch caching algorithms, using a probing algorithm that is asymptotically optimal in terms of probing overhead. We demonstrate cases where routing and scheduling optimizations based on Tango improves the rule installation time by up to 70% in our hardware switch testbed.

The current LTE network architecture is organized into very large regions, each having a core network and a radio access network. The core network contains an Internet edge comprised of packet data network gateways (PGWs). The radio network consists of only base stations. There are minimal interactions among regions other than interference management at the edge. The current architecture has several problems. First, mobile application performance is seriously impacted by the lack of Internet egress points per region. Second, the continued exponential growth of mobile traffic puts tremendous pressure on the scalability of PGWs. Third, the fast growth of signaling traffic known as the signaling storm problem poses a major challenge to the scalability of the control plane. To address these problems, we present SoftMoW, a recursive and reconfigurable cellular WAN architecture that supports seamlessly inter-connected core networks, reconfigurable control plane, and global optimization.

With the continued exponential growth of mobile traffic and the rise of diverse applications, the current LTE radio access network (RAN) architecture of cellular operators face mounting challenges. Current RAN suffers from insufficient radio resource coordination, inefficient infrastructure utilization, and inflexible data paths. We present the high level design of PRAN, which centralizes base stations' L1/L2 processing into a cluster of commodity servers. PRAN uses a flexible data path model to support new protocols; multiple base stations' L1/L2 processing tasks are scheduled on servers with performance guarantees; and a RAN scheduler coordinates the allocation of shared radio resources between operators and base stations. Our evaluation shows the feasibility of fast data path control and efficiency of resource pooling (a potential for a 30× reduction on resources).

Despite the rapid growth of next-generation cellular networks, researchers and end-users today have limited visibility into the performance and problems of these networks. As LTE deployments move towards femto and pico cells, even operators struggle to fully understand the propagation and interference patterns affecting their service, particularly indoors. This paper introduces LTEye, the first open platform to monitor and analyze LTE radio performance at a fine temporal and spatial granularity. LTEye accesses the LTE PHY layer without requiring private user information or provider support. It provides deep insights into the PHY-layer protocols deployed in these networks. LTEye's analytics enable researchers and policy makers to uncover serious deficiencies in these networks due to inefficient spectrum utilization and inter-cell interference. In addition, LTEye extends synthetic aperture radar (SAR), widely used for radar and backscatter signals, to operate over cellular signals. This enables businesses and end-users to localize mobile users and capture the distribution of LTE performance across spatial locations in their facility. As a result, they can diagnose problems and better plan deployment of repeaters or femto cells. We implement LTEye on USRP software radios, and present empirical insights and analytics from multiple AT&T and Verizon base stations in our locality.

Wireless security has been an active research area since the last decade. A lot of studies of wireless security use cryptographic tools, but traditional cryptographic tools are normally based on computational assumptions, which may turn out to be invalid in the future. Consequently, it is very desirable to build cryptographic tools that do not rely on computational assumptions. In this paper, we focus on a crucial cryptographic tool, namely 1-out-of-2 oblivious transfer. This tool plays a central role in cryptography because we can build a cryptographic protocol for any polynomial-time computable function using this tool. We present a novel 1-out-of-2 oblivious transfer protocol based on wireless channel characteristics, which does not rely on any computational assumption. We also illustrate the potential broad applications of this protocol by giving two applications, one on private communications and the other on privacy preserving password verification. We have fully implemented this protocol on wireless devices and conducted experiments in real environments to evaluate the protocol. Our experimental results demonstrate that it has reasonable efficiency.

Cellular core networks suffer from inflexible and expensive equipment, as well as from complex control-plane protocols. To address these challenges, we present SoftCell, a scalable architecture that supports fine-grained policies for mobile devices in cellular core networks, using commodity switches and servers. SoftCell enables operators to realize high-level service policies that direct traffic through sequences of middleboxes based on subscriber attributes and applications. To minimize the size of the forwarding tables, SoftCell aggregates traffic along multiple dimensions---the service policy, the base station, and the mobile device---at different switches in the network. Since most traffic originates from mobile devices, SoftCell performs fine-grained packet classification at the access switches, next to the base stations, where software switches can easily handle the state and bandwidth requirements. SoftCell guarantees that packets belonging to the same connection traverse the same sequence of middleboxes in both directions, even in the presence of mobility. We demonstrate that SoftCell improves the scalability and flexibility of cellular core networks by analyzing real LTE workloads, performing micro-benchmarks on our prototype controller as well as large-scale simulations.

Like their public counterpart such as Facebook and Twitter, enterprise social networks are poised to revolutionize how people interact in the workplace. There is a pressing need to understand how people are using these social networks. Unlike the public social networks like Facebook or Twitter which are normally characterized using the social graph or the interaction graph, enterprise social networks are also governed by an organization graph. Based on a six month dataset collected from May through October 2011 of a large enterprise social network, we study the characteristics of activities of its enterprise social network. We observe that the user attributes in the organization graph such as geographic location (eg. country) and his/her rank in the company hierarchy have a significant impact on how the user uses the social network and how user interacts with each other. We then build formal statistical models of user interaction graphs in enterprise social network and quantify effects of user attributes from organization graphs on these interactions. Furthermore, as the enterprise social network medium bring users from diverse locations and social status forming ad-hoc communities, our statistical model can be further enhanced by including these ad-hoc communities.

The emergence of new capabilities such as virtualization and elastic (private or public) cloud computing infrastructures has made it possible to deploy multiple applications, on demand, on the same cloud infrastructure. A major challenge to achieve this possibility, however, is that modern applications are typically distributed, structured systems that include not only computational and storage entities, but also policy entities (e.g., load balancers, firewalls, intrusion prevention boxes). Deploying applications on a cloud infrastructure without the policy entities may introduce substantial policy violations and/or security holes. In this paper, we present PACE: the first systematic framework for Policy-Aware Application Cloud Embedding. We precisely define the policy-aware, cloud application embedding problem, study its complexity and introduce simple, efficient, online primal-dual algorithms to embed applications in cloud data centers. We conduct evaluations using data from a real, large campus network and a realistic data center topology to evaluate the feasibility and performance of PACE. We show that deployment in a cloud without considering in-network policies may lead to a large number of policy violations (e.g., using tree routing as a way to enforce in-network policies may observe up to 91% policy violations). We also show that our embedding algorithms are very efficient by comparing with a good online fractional embedding algorithm.

Spectrum sensing, the task of discovering spectrum usage at a given location, is a fundamental problem in dynamic spectrum access networks. While sensing in narrow spectrum bands is well studied in previous work, wideband spectrum sensing is challenging since a wideband radio is generally too expensive and power consuming for mobile devices. Sequential scan, on the other hand, can be very slow if the wide spectrum band contains many narrow channels. In this paper, we propose an analog-filter based spectrum sensing technique, which is much faster than sequential scan and much cheaper than using a wideband radio. The key insight is that, if the sum of energy on a contiguous band is low, we can conclude that all channels in this band are clear with just one measurement. Based on this insight, we design an intelligent search algorithm to minimize the number of total measurements. We prove that the algorithm has the same asymptotic complexity as compressed sensing while our design is much simpler and easily implementable in the real hardware. We show the availability of our technique using hardware devices that include analog filters and analog energy detectors. Our extensive evaluation using real TV “white space” signals shows the effectiveness of our technique.

Multi-user multiple-input multiple-output theory predicts manyfold capacity gains by leveraging many antennas on wireless base stations to serve multiple clients simultaneously through multi-user beamforming (MUBF). However, realizing a base station with a large number antennas is non-trivial, and has yet to be achieved in the real-world. We present the design, realization, and evaluation of Argos, the first reported base station architecture that is capable of serving many terminals simultaneously through MUBF with a large number of antennas (M >> 10). Designed for extreme flexibility and scalability, Argos exploits hierarchical and modular design principles, properly partitions baseband processing, and holistically considers real-time requirements of MUBF. Argos employs a novel, completely distributed, beamforming technique, as well as an internal calibration procedure to enable implicit beamforming with channel estimation cost independent of the number of base station antennas. We report an Argos prototype with 64 antennas and capable of serving 15 clients simultaneously. We experimentally demonstrate that by scaling from 1 to 64 antennas the prototype can achieve up to 6.7 fold capacity gains while using a mere 1/64th of the transmission power.

Multiparty interactive network applications such as teleconferencing, network gaming, and online trading are gaining popularity. In addition to end-to-end latency bounds, these applications require that the delay difference among multiple clients of the service is minimized for a good interactive experience. We propose a Latency EQualization (LEQ) service, which equalizes the perceived latency for all clients participating in an interactive network application. To effectively implement the proposed LEQ service, network support is essential. The LEQ architecture uses a few routers in the network as hubs to redirect packets of interactive applications along paths with similar end-to-end delay. We first formulate the hub selection problem, prove its NP-hardness, and provide a greedy algorithm to solve it. Through extensive simulations, we show that our LEQ architecture significantly reduces delay difference under different optimization criteria that allow or do not allow compromising the per-user end-to-end delay. Our LEQ service is incrementally deployable in today's networks, requiring just software modifications to edge routers.

Wireless security has been an active research area since the last decade. A lot of studies of wireless security use cryptographic tools, but traditional cryptographic tools are normally based on computational assumptions, which may turn out to be invalid in the future. Consequently, it is very desirable to build cryptographic tools that do not rely on computational assumptions. In this paper, we focus on a crucial cryptographic tool, namely 1-out-of-2 oblivious transfer. This tool plays a central role in cryptography because we can build a cryptographic protocol for any polynomial-time computable function using this tool. We present a novel 1-out-of-2 oblivious transfer protocol based on wireless channel characteristics, which does not rely on any computational assumption. We also illustrate the potential broad applications of this protocol by giving an application on private communications. We have fully implemented this protocol on wireless devices and conducted experiments in real environments to evaluate the protocol and its application to private communications. Our experimental results demonstrate that it has reasonable efficiency.

Existing media providers such as YouTube and Hulu deliver videos by turning it into a progressive download. This can result in frequent video freezes under varying network dynamics. In this paper, we present CloudStream: a cloud-based video proxy that can deliver high-quality streaming videos by transcoding the original video in real time to a scalable codec which allows streaming adaptation to network dynamics. The key is a multi-level transcoding parallelization framework with two mapping options (Hallsh-based Mapping and Lateness-first Mapping) that optimize transcoding speed and reduce the transcoding jitters while preserving the encoded video quality. We evaluate the performance of CloudStream on our campus cloud testbed.

This paper proposes and studies a system, called CloudFlex, which transparently taps cloud resources to serve application requests that exceed capacity of internal infrastructure. CloudFlex operates as a feedback control system with two key interacting components: load balancer and controller. We focus on operational optimality and stability of the system, highlight the tradeoffs between cost and responsiveness, and address important design considerations such as choke point detection that are critical in avoiding pathological system operations. For evaluation, we develop a prototype of CloudFlex on our testbed comprising servers of our enterprise data center and Amazon EC2 instances.

As a key approach to securing large networks, existing anomaly detection techniques focus primarily on network traffic data. However, the sheer volume of such data often renders detailed analysis very expensive and reduces the effectiveness of these tools. In this paper, we propose a light-weight anomaly detection approach based on unproductive DNS traffic, namely, the failed DNS queries, with a novel tool - DNS failure graphs. A DNS failure graph captures the interactions between hosts and failed domain names. We apply a graph decomposition algorithm based on the tri-nonnegative matrix factorization technique to iteratively extract coherent co-clusters (dense subgraphs) from DNS failure graphs. By analyzing the co-clusters in the daily DNS failure graphs from a 3-month DNS trace captured at a large campus network, we find these co-clusters represent a variety of anomalous activities, e.g., spamming, trojans, bots, etc.. In addition, these activities often exhibit distinguishable subgraph structures. By exploring the temporal properties of the co-clusters, we show our method can identify new anomalies that likely correspond to unreported domain-flux bots.

Quantiles are very useful in characterizing the data distribution of an evolving dataset in the process of data mining or network monitoring. The method of Stochastic Approximation (SA) tracks quantiles online by incrementally deriving and updating local approximations of the underly distribution function at the quantiles of interest. In this paper, we propose a generalization of the SA method for quantile estimation that allows not only data insertions, but also dynamic data operations such as deletions and updates.

Physical layer techniques have come a long way and can achieve very close to Shannon capacity for point-to-pint links. It is apparent that, to further improve network capacity significantly, we have to resort to concurrent transmissions. Many concurrent transmission techniques (e.g., zero forcing, interference alignment and distributed MIMO) are proposed in which multiple senders jointly encode signals to multiple receivers so that interference is aligned and each receiver is able to decode its desired information. In this paper, we investigate the constraints and challenges of using interference alignment. Our main contribution is conducting the first systematic investigation on the key issue of identifying opportunities for interference alignment. We identify diverse, novel scenarios for using interference alignment. We show that identifying opportunities for interference alignment in the general case is computational challenging. However, we also present a promising, distributed algorithm for identifying a wide range of opportunities for interference alignment using a unifying framework based on the degree of freedom. Our second contribution is evaluating key practical implementation issues.

Collisions in overlapping channels can be a major problem in the deployment of high-speed OFDM networks. In this paper, we present Remap, a simple, novel paradigm for handling collisions in overlapping OFDM channels. Remap introduces a novel concept of retransmission permutation that permutes the bit-to-subcarrier assignment after each transmission, departing from the traditional, simply-repeat paradigm. Remap is simple to implement and able to exploit collision-free subcarriers to decode frames despite successive collisions in overlapping channels. We apply Remap to 802.11g to demonstrate that the diversity created by remapped frames can substantially improve decoding efficiency and improve wireless throughput. We implement our technique in software radio and demonstrate that it has potential to be deployed with simple software and firmware updates.

Understanding the aggregate behavior of network host connectivities is important for network monitoring and traffic engineering. One characterization of such an aggregate behavior is the host distributions of distinct communicating peers or flows. For example, during the worm outbreak, the port scanning activities would cause many hosts with increasing number of (one-way) peers (or flows), and hence a change in the host distributions of distinct communicating peers or flows. In this paper, we develop an efficient streaming algorithm for tracking these host distributions of distinct elements, also called cardinality distributions, for a high speed network with a large number of hosts. Our approach utilizes the continuous Flajolet-Martin sketches, which is the minimal order statistics of hashed values, as a compact data summary and develops maximum likelihood estimates of these distributions. By leveraging the aggregation of many hosts, we are able to obtain very accurate estimates of the cardinality distributions by maintaining a compact statistical summary that is as small as one number (at most 32 bits) per host. Extensive experimental studies are carried out to demonstrate their excellent performance.

We present muNet, a wireless mesh network design and implementation to harness the multiuser capacity of wireless channels. Traditionally, media access control is designed to schedule one transmission between one sender and one receiver without interference at any given time. However, this design is suboptimal in terms of achieving the multiuser capacity of multi-access wireless channels. In muNet, we implement effective physical layer techniques called superposition coding and successive interference cancellation to enable simultaneous unicast transmissions from a single transmitter to multiple receivers as well as from multiple transmitters to a single receiver. We design the first practical MAC protocol that leverages such a physical layer and exposes the multiuser capacity to upper layers. We also present a simple, effective routing protocol that increases simultaneous transmission opportunities for the MAC layer. A proof-of-concept muNet is implemented on the GNU radio platform. Measurements on the implementation shows that the throughput gains of muNet are significant (up to 93%).

A major barrier for the adoption of wireless mesh networks is severe limits on throughput. Many in-network packet mixing techniques at the network layer [1], [2], [3] as well as the physical layer [4], [5], [6] have been shown to substantially improve throughput. However, the optimal mixing algorithm that maximizes throughput is still unknown. We propose iPack, an algorithm for in-network generation of composite packets that integrates coding at two different layers of the protocol stack: XOR-based network coding and physical layer superposition coding. Using extensive simulations, we find that the throughput gain of the joint coding iPack algorithm is 30% more than the better performer of network coding and superposition coding in a wide range of scenarios, and automatically takes advantage of the best available coding opportunities. In a typical wireless mesh network when more traffic is between the clients and access points, the average throughput improvement of iPack, our joint optimization scheduler, can be 324%, while there can be little gain (less than 10%) if network coding alone is used. We also validate our results by implementing iPack on a small-scale testbed based on GNU Radio.

IP network mobility is emerging as a major paradigm for providing continuous Internet access while a set of users are on the move in a transportation system. The intense interest on its support has led to the establishment of the NEMO IETF working group and a test-deployment by a major airline equipment vendor - Boeing - on major airline routes. However, the previously proposed solutions are either inefficient or may cause instability to the global Internet. We propose WINMO, a simple, systematic, novel solution for wide-area IP network mobility using techniques including route aggregation, scoped update propagation, and packet mobility states. Our solution provides efficient routing when users travel both across autonomous systems (ASes) and within a single AS, generates minimal global routing overhead to prevent global instability, ensures good location privacy, and helps to defend against denial-of-service attacks. Furthermore, our basic scheme (without packet mobility state) is transparent to both clients and servers. Our extensive evaluations demonstrate the effectiveness of our mobility solution.

In multi-rate wireless LANs, throughput-based fair bandwidth allocation can lead to drastically reduced aggregate throughput. To balance aggregate throughput while serving users in a fair manner, proportional fair or time-based fair scheduling has been proposed to apply at each access point (AP). However, since a realistic deployment of wireless LANs can consist of a network of APs, this paper considers proportional fairness in this much wider setting. Our technique is to intelligently associate users with APs to achieve optimal proportional fairness in a network of APs. We propose two approximation algorithms for periodical offline optimization. Our algorithms are the first approximation algorithms in the literature with a tight worst-case guarantee for the NP-hard problem. Our simulation results demonstrate that our algorithms can obtain an aggregate throughput which can be as much as 2.3 times more than that of the max-min fair allocation in 802.11b. While maintaining aggregate throughput, our approximation algorithms outperform the default user-AP association method in the 802.11b standard significantly in terms of fairness.

User-contributed wireless mesh networks are a disruptive technology that may fundamentally change the economics of edge network access and bring the benefits of a computer network infrastructure to local communities at low cost, anywhere in the world. To achieve high throughput despite highly unpredictable and lossy wireless channels, it is essential that such networks take advantage of transmission opportunities wherever they emerge. However, as opportunistic routing departs from the traditional but less effective deterministic, shortest-path based routing, user nodes in such networks may have less incentive to follow protocols and contribute. In this paper, we present the first routing protocols in which it is incentive-compatible for each user node to honestly participate in the routing despite opportunistic transmissions. We not only rigorously prove the properties of our protocols but also thoroughly evaluate a complete implementation of our protocols. Experiments show that there is a 5.8%-58.0% gain in throughput when compared with an opportunistic routing protocol that does not provide incentives and users can act selfishly.

Information on network host connectivity patterns are important for network monitoring and traffic engineering. In this paper, an efficient streaming algorithm is proposed to estimate cardinality distributions including connectivity distributions, e.g. percent of hosts with any given number of distinct communicating peers or flows.

Power efficient operation is very critical in energy constrained multi-hop wireless networks. One important technique is to intelligently assign transmission powers to nodes while maintaining network connectivity. Previous work has focused on assigning a single transmission power to each node. This often leads to "power imbalance", where some nodes use much more power than other nodes. This can reduce network lifetime. In this paper, we investigate the problem of two power assignments where nodes alternate the use of these assigned powers. We rigorously formulate the problem of two power assignment under the constraint that the network connectivity is maintained. The objective here is to minimize the maximum average power used by the nodes. We show that, in general, the problem is not just NP-hard but also hard to approximate. We then propose a distributed localized heuristic to compute the two power assignments. We perform extensive simulations to show that the algorithm can reduce the average power significantly when compared with algorithms that assign a single power. By assuming some properties on radio propagation, we also present a centralized algorithm with bounded worst case guarantees for the two power assignment problem.

Multiple input multiple output (MIMO) antennas use sophisticated physical layer techniques to provide significant benefits over conventional antenna technology. Multiple independent data streams can be sent over the MIMO antenna elements. MIMO link can also suppress interference from neighboring links as long as the total useful streams and interfering streams are no greater than the number of receiving antenna elements. For these reasons MIMO antennas are increasingly being considered for use in interference limited wireless mesh networks and have been adopted by WLAN and WIMAX standards. However, the benefits of the MIMO technology in improving network performance are limited unless the higher layer protocols also exploit these capabilities. In this paper we are interested in characterizing the benefits of cross-layer optimizations in interference limited wireless mesh networks with MIMO links. We formulate a framework where data routing at the protocol layer, link scheduling at the MAC layer and stream control at the physical layer can be jointly optimized for throughput maximization in the presence of interference. We then develop an efficient algorithm to solve the resulting throughput optimization problem subject to fairness constraints.

Broadcast operation, which disseminates information network-wide, is very important in multi-hop wireless networks. Due to the broadcast nature of wireless media, not all nodes need to transmit in order for the message to reach every node. Previous work on broadcast support can be classified as probabilistic (each node rebroadcasts a packet with a given probability) or deterministic approaches (nodes pre-select a few neighbors for rebroadcasting). In this paper, we show how network-coding can be applied to a deterministic broadcast approaches, resulting in significant reductions in the number of transmissions in the network. We propose two algorithms, that rely only on local two-hop topology information and makes extensive use of opportunistic listening to reduce the number of transmissions: 1) a simple XOR-based coding algorithm that provides up to 45% gains compared to a non-coding approach and 2) a Reed-Solomon based coding algorithm that determines the optimal coding gain achievable for a coding algorithm that relies only on local information, with gains up to 61% in our simulations. We also show that our coding-based deterministic approach outperforms the coding-based probabilistic approach presented in (C. Fragouli et al, 2006).

A major barrier for the adoption of wireless mesh networks is severe limits on throughput. In this paper, we apply superposition coding to substantially improve network capacity of large, dense wireless mesh networks. Superposition coding is a physical layer technique that allows a transmitter to simultaneously send independent packets to multiple receivers. While superposition coding has been studied extensively by the physical layer community, we present the first design of practical and effective MAC protocols to take advantage of superposition coding in wireless mesh networks. Extensive evaluations show that superposition coding can be a practical method to increase the throughput of large, dense wireless mesh networks. Specifically, in a mesh network with 2 to 64 active receivers and one gateway, we show that our system can increase throughput up to 154%, with average gain ranging from 10% to 21%. When there are multiple gateways forming a mesh network, our system gains up to 98%, with average gain ranging from 24% to 46%. These results clearly demonstrate the potential benefits of our system. We also present results from an implementation of superposition coding using GNU Radio.

In third-generation (3G) wireless data networks, providing service to low data-rate users is required for maintaining fairness, but at the cost of reducing the cell's aggregate throughput. In this paper, we propose the unified cellular and ad hoc network (UCAN) architecture for enhancing cell throughput while maintaining fairness. In UCAN, a mobile client has both 3G interface and IEEE 802.11 -based peer-to-peer links. The 3G base station forwards packets for destination clients with poor channel quality to proxy clients with better channel quality. The proxy clients then use an ad hoc network composed of other mobile clients and IEEE 802.11 wireless links to forward the packets to the appropriate destinations, thereby improving cell throughput. We refine the 3G base station scheduling algorithm so that the throughput gains are distributed in proportion to users' average channel rates, thereby maintaining fairness. With the UCAN architecture in place, we propose novel greedy and on-demand protocols for proxy discovery and ad hoc routing that explicitly leverage the existence of the 3G infrastructure to reduce complexity and improve reliability. We further propose secure crediting mechanisms to motivate users that are not actively receiving to participate in relaying packets for others. Through both analysis and extensive simulations with HDR and IEEE 802.11b, we show that the UCAN architecture can increase individual user's throughput by more than 100 percent and the aggregate throughput of the HDR downlink by up to 50 percent.

Abstract — With the increasing commercial interest in supporting voice and multimedia services over the IP network there is a need for bandwidth guaranteed services. For example, guaranteeing the peak demand of VoIP traffic entails high costs in terms of bandwidth reservation requirements. To effectively make use of the reserved peak bandwidth, it is imperative that this bandwidth is shared with best effort data traffic during non peak periods. In this paper, we formulate this bandwidth sharing network design problem. Our goal is to minimize the total cost of bandwidth reservation while satisfying (1) the peak demand for real time traffic, and (2) the average demand of both real time and best effort data traffic. We show that, the problem is polynomially solvable if we do not restrict the number of paths used. It is strongly NP-hard if we use only one path between any pair of nodes. We present a simple 2-approximation algorithm and through simulation studies show that this algorithm can be further improved using a local search heuristic. Our simulation results also show that sharing significantly reduces the total bandwidth reservation costs and our local search heuristic can find a solution which is very close to or equal to the optimal in most cases.

Multihop infrastructure wireless mesh networks offer increased reliability, coverage, and reduced equipment costs over their single-hop counterpart, wireless local area networks. Equipping wireless routers with multiple radios further improves the capacity by transmitting over multiple radios simultaneously using orthogonal channels. Efficient channel assignment and routing is essential for throughput optimization of mesh clients. Efficient channel assignment schemes can greatly relieve the interference effect of close-by transmissions; effective routing schemes can alleviate potential congestion on any gateways to the Internet, thereby improving per-client throughput. Unlike previous heuristic approaches, we mathematically formulate the joint channel assignment and routing problem, taking into account the interference constraints, the number of channels in the network, and the number of radios available at each mesh router. We then use this formulation to develop a solution for our problem that optimizes the overall network throughput subject to fairness constraints on allocation of scarce wireless capacity among mobile clients. We show that the performance of our algorithms is within a constant factor of that of any optimal algorithm for the joint channel assignment and routing problem. Our evaluation demonstrates that our algorithm can effectively exploit the increased number of channels and radios, and it performs much better than the theoretical worst case bounds

In third-generation (3G) wireless data networks, repeated requests for popular data items can exacerbate the already scarce wireless spectrum. In this paper, we propose an architectural and protocol framework that allows 3G service providers to host efficient content distribution services. We offload the spectrum intensive task of content distribution to an ad hoc network. Less mobile users (resident subscribers) are provided incentives to cache popular data items, while mobile users (transit subscribers) access this data from resident subscribers through the ad hoc network. Since the participants of this data distribution network act as selfish agents, they may collude to maximize their individual payoff. Our proposed protocol discourages potential collusion scenarios. In this architecture, the goal (social function) of the 3G service provider is to have the selfishly motivated resident subscribers service as many data requests as possible. However, the choice of which set of items to cache is left to the individual user. The caching activity among the different users can be modeled as a market sharing game. In this work, we study the Nash equilibria of market sharing games and the performance of such equilibria in terms of a social function. These games are a special case of congestion games that have been studied in the economics literature. In particular, pure strategy Nash equilibria for this set of games exist. We give a polynomial-time algorithm to find a pure strategy Nash equilibrium for a special case, while it is NP-hard to do so in the general case. As for the performance of Nash equilibria, we show that the price of anarchy-the worst case ratio between the social function at any Nash equilibrium and at the social optimum-can be upper bounded by a factor of 2. When the popularity follows a Zipf distribution, the price of anarchy is bounded by 1.45 in the special case where caching any item has a positive reward for all players. We prove that the selfish behavior of computationally bounded agents converges to an approximate Nash equilibrium in a finite number of improvements. Furthermore, we prove that, after each agent computes its response function once using a constant factor approximation algorithm, the outcome of the game is within a factor of O(logn) of the optimal social value, where n is the number of agents. Our simulation scenarios show that the price of anarchy is 30% better than that of the worst case analysis and that the system quickly (1 or 2 steps) converges to a Nash equilibrium.

In third generation (3G) wireless data networks, multicast throughput decreases with the increase in multicast group size, since a conservative strategy for the base station is to use the lowest data rate of all the receivers so that the receiver with the worst downlink channel condition can decode the transmission correctly. This paper proposes ICAM, integrated cellular and ad hoc multicast, to increase 3G multicast throughput through opportunistic use of ad hoc relays. In ICAM, a 3G base station delivers packets to proxy mobile devices with better 3G channel quality. The proxy then forwards the packets to the receivers through an IEEE 802.11-based ad hoc network. In this paper, we first propose a localized greedy algorithm that discovers for each multicast receiver the proxy with the highest 3G downlink channel rate. We discover that due to capacity limitations and interference of the ad hoc relay network, maximizing the 3G downlink data rate of each multicast receiver's proxy does not lead to maximum throughput for the multicast group. We then show that the optimal ICAM problem is NP-hard, and derive a polynomial-time 4-approximation algorithm for the construction of the multicast forest. This bound holds when the underlying wireless MAC supports broadcast or unicast, single rate or multiple rates (4(1 + /spl isin/) approximation scheme for the latter), and even when there are multiple simultaneous multicast sessions. Through both analysis and simulations, we show that our algorithms achieve throughput gains up to 840 percent for 3G downlink multicast with modest overhead on the 3G uplink.

We present a general model of interdomain route selection to study interdomain traffic engineering. In this model, the routing of multiple destinations can be coordinated. Thus the model can capture general traffic engineering behaviors such as load balancing and link capacity constraints. We first identify potential routing instability and inefficiency of interdomain traffic engineering. We then derive a sufficient condition to guarantee convergence. We also show that the constraints on local policies imposed by business considerations in the Internet can guarantee stability without global coordination. Using realistic Internet topology, we evaluate the extent to which routing instability of interdomain traffic engineering can happen when the constraints are violated.

The recent discovery of instability caused by the interaction of local routing policies of multiple ASes has led to extensive research on the subject. However, previous studies analyze stability under a specific route selection algorithm. In this paper, instead of studying a specific route selection algorithm, we study a general class of route selection algorithms which we call rational route selection algorithms. We present a sufficient condition to guarantee routing convergence in a heterogeneous network where each AS runs any rational route selection algorithm. Applying our general results, we study the potential instability of a network where the preference of an AS depends on not only its egress routes to the destinations but also its inbound traffic patterns (i.e., the distribution of incoming traffic from its neighbors). We show that there exist networks which will have persistent route oscillations even when the ASes strictly follow the constraints imposed by business considerations, and adopt any rational route selection algorithms.

Recent advances in Internet measurement tools have made it possible to locate bottleneck links that constrain the available bandwidth of Internet paths. In this paper, we provide a detailed study of Internet path bottlenecks. We focus on the following four aspects: the persistence of bottleneck location, the sharing of bottlenecks among destination clusters, the packet loss and queueing delay of bottleneck links, and the relationship with router and link properties, including router CPU load, router memory load, link traffic load, and link capacity. We find that 20% - 30% of the source-destination pairs in our measurement have a persistent bottleneck; fewer than 10% of the destinations in a prefix cluster share a bottleneck more than half of the time; 60% of the bottlenecks on lossy paths can be correlated with a loss point no more than 2 hops away; and bottlenecks can be clearly correlated with link load, while presenting no strong relationship with link capacity, router CPU and memory load.

Multi-hop infrastructure wireless mesh networks offer increased reliability, coverage and reduced equipment costs over their single-hop counterpart, wireless LANs. Equipping wireless routers with multiple radios further improves the capacity by transmitting over multiple radios simultaneously using orthogonal channels. Efficient channel assignment and routing is essential for throughput optimization of mesh clients. Efficient channel assignment schemes can greatly relieve the interference effect of close-by transmissions; effective routing schemes can alleviate potential congestion on any gateways to the Internet, thereby improving per-client throughput. Unlike previous heuristic approaches, we mathematically formulate the joint channel assignment and routing problem, taking into account the interference constraints, the number of channels in the network and the number of radios available at each mesh router. We then use this formulation to develop a solution for our problem that optimizes the overall network throughput subject to fairness constraints on allocation of scarce wireless capacity among mobile clients. We show that the performance of our algorithms is within a constant factor of that of any optimal algorithm for the joint channel assignment and routing problem. Our evaluation demonstrates that our algorithm can effectively exploit the increased number of channels and radios, and it performs much better than the theoretical worst case bounds.

In many applications, wireless ad-hoc networks are formed by devices belonging to independent users. Therefore, a challenging problem is how to provide incentives to stimulate cooperation. In this paper, we study ad-hoc games---the routing and packet forwarding games in wireless ad-hoc networks. Unlike previous work which focuses either on routing or on forwarding, this paper investigates both routing and forwarding. We first uncover an impossibility result---there does not exist a protocol such that following the protocol to always forward others' traffic is a dominant action. Then we define a novel solution concept called cooperation-optimal protocols. We present Corsac, a cooperation-optimal protocol consisting of a routing protocol and a forwarding protocol. The routing protocol of Corsac integrates VCG with a novel cryptographic technique to address the challenge in wireless ad-hoc networks that a link's cost (ie, its type) is determined by two nodes together. Corsac also applies efficient cryptographic techniques to design a forwarding protocol to enforce the routing decision, such that fulfilling the routing decision is the optimal action of each node in the sense that it brings the maximum utility to the node. Additionally, we extend our framework to a practical radio propagation model where a transmission is successful with a probability. We evaluate our protocols using simulations. Our evaluations demonstrate that our protocols provide incentives for nodes to forward packets.

In this paper, we consider the multicast throughput optimization problem in multi-hop wireless networks. Given a source, and a set of receivers, we would like to find the set of multicast trees and a schedule such that the rate that the source can multicast to the receivers is maximized. We consider two transmission models: broadcast and unicast. In the broadcast model, a transmission is received by multiple downstream nodes in a multicast tree. In the unicast model, a separate transmission has to be sent to each downstream node. We consider the fundamental constraint that a node can not be involved in multiple communications at the same time. We consider two multicast models: a single multicast tree per session and multiple multicast tree per session. In the single multicast tree case, (1) for the unicast model, we show that the problem is NP-hard and it is not approximable to a factor better than 1.5; we then give a 1.5-approximation algorithm if all links have the same data rate, a 5-approximation algorithm if all nodes have the same transmission power and a 24-approximation algorithm for a realistic heterogeneous ad hoc network where nodes can have different transmission power. (2) for the broadcast model, we show that the problem is NP-hard and it is not approximable to a factor better than 2; we then give a simple 2-approximation algorithm to find the multicast tree and the transmission schedule. In the multiple multicast tree case, (1) for the unicast model, we show that the problem is APX-hard, and give a 1.5Ρ-approximation where Ρ is the best approximation ratio of the minimal cost Steiner tree problem; (2) for the broadcast model, our results indicate that the problem is hard, may not be approximable within a factor better than log(n) where n is the number of multicast receivers. Our evaluation shows that the throughput achieved by our algorithms is much better than both the throughput achieved by using pruned shortest path tree and by using optimal unicast.

The emerging multiprotocol label switching (MPLS) networks enable network service providers to route bandwidth guaranteed paths between customer sites. This basic label switched path (LSP) routing is often enhanced using restoration routing which sets up alternate LSPs to guarantee uninterrupted connectivity in case network links or nodes along primary path fail. We address the problem of distributed routing of restoration paths, which can be defined as follows: given a request for a bandwidth guaranteed LSP between two nodes, find a primary LSP, and a set of backup LSPs that protect the links along the primary LSP. A routing algorithm that computes these paths must optimize the restoration latency and the amount of bandwidth used. We introduce the concept of "backtracking" to bound the restoration latency. We consider three different cases characterized by a parameter called backtracking distance D: 1) no backtracking (D=0); 2) limited backtracking (D=k); and 3) unlimited backtracking (D=/spl infin/). We use a link cost model that captures bandwidth sharing among links using various types of aggregate link-state information. We first show that joint optimization of primary and backup paths is NP-hard in all cases. We then consider algorithms that compute primary and backup paths in two separate steps. Using link cost metrics that capture bandwidth sharing, we devise heuristics for each case. Our simulation study shows that these algorithms offer a way to tradeoff bandwidth to meet a range of restoration latency requirements.

The topology of a wireless multi-hop network can be controlled by varying the transmission power at each node. In this paper, we give a detailed analysis of a cone-based distributed topology-control (CBTC) algorithm. This algorithm does not assume that nodes have GPS information available; rather it depends only on directional information. Roughly speaking, the basic idea of the algorithm is that a node u transmits with the minimum power p/sub u,/spl alpha// required to ensure that in every cone of degree /spl alpha/ around u, there is some node that u can reach with power p/sub u,/spl alpha//. We show that taking /spl alpha/=5/spl pi//6 is a necessary and sufficient condition to guarantee that network connectivity is preserved. More precisely, if there is a path from s to t when every node communicates at maximum power then, if /spl alpha//spl les/5/spl pi//6, there is still a path in the smallest symmetric graph G/sub /spl alpha// containing all edges (u,v) such that u can communicate with v using power p/sub u,/spl alpha//. On the other hand, if /spl alpha/>5/spl pi//6, connectivity is not necessarily preserved. We also propose a set of optimizations that further reduce power consumption and prove that they retain network connectivity. Dynamic reconfiguration in the presence of failures and mobility is also discussed. Simulation results are presented to demonstrate the effectiveness of the algorithm and the optimizations.

Evolving monitoring infrastructure in response to network growth is a critical aspect of network management. Previous work in network management primarily focused on optimizing monitoring systems for static networks. In this paper, we address the problem of optimally upgrading the existing monitoring infrastructure as the network evolves. The problem formulation presented here captures the trade off between adding new monitoring resources vs. disrupting the existing infrastructure. We show that this problem is NP hard and not approximable within a factor better than n/sup /spl epsi// in the general case and no better than log(n) when only shortest paths are considered. We develop a heuristic algorithm and evaluate its performance using simulated network evolution scenarios. We show that in spite of not allowing poller relocation, our adaptive online algorithm has comparable performance to that of the offline algorithm where such constraint does not exist.

Per-flow traffic measurement is critical for usage accounting, traffic engineering, and anomaly detection. Previous methodologies are either based on random sampling (e.g., Cisco's NetFlow), which is inaccurate, or only account for the "elephants". We introduce a novel technique for measuring per-flow traffic approximately, for all flows regardless of their sizes, at very high-speed (say, OC768). The core of this technique is a novel data structure called space code bloom filter (SCBF). A SCBF is an approximate representation of a multiset; each element in this multiset is a traffic flow and its multiplicity is the number of packets in the flow. The multiplicity of an element in the multiset represented by SCBF can be estimated through either of two mechanisms-maximum likelihood estimation (MLE) or mean value estimation (MVE). Through parameter tuning, SCBF allows for graceful tradeoff between measurement accuracy and computational and storage complexity. SCBF also contributes to the foundation of data streaming by introducing a new paradigm called blind streaming. We evaluate the performance of SCBF through mathematical analysis and through experiments on packet traces gathered from a tier-1 ISP backbone. Our results demonstrate that SCBF achieves reasonable measurement accuracy with very low storage and computational complexity

Recent studies on operational wireless LANs (WLANs) have shown that user load is often unevenly distributed among wireless access points (APs). This unbalanced load results in unfair bandwidth allocation among users. We observe that the unbalanced load and unfair bandwidth allocation can be greatly alleviated by intelligently associating users to APs, termed association control, rather than having users greedily associate APs of best received signal strength.In this study, we present an efficient algorithmic solution to determine the user-AP associations that ensure max-min fair bandwidth allocation. We provide a rigorous formulation of the association control problem that considers bandwidth constraints of both the wireless and backhaul links. Our formulation indicates the strong correlation between fairness and load balancing, which enables us to use load balancing techniques for obtaining near optimal max-min fair bandwidth allocation. Since this problem is NP-hard, we present algorithms that achieve a constant-factor approximate max-min fair bandwidth allocation. First, we calculate a fractional load balancing solution, where users can be associated with multiple APs simultaneously. This solution guarantees the fairest bandwidth allocation in terms of max-min fairness. Then, by utilizing a rounding method we obtain an efficient integral association. In particular, we provide a 2-approximation algorithm for unweighted greedy users and a 3-approximation algorithm for weighted and bounded-demand users. In addition to bandwidth fairness, we also consider time fairness and we show it can be solved optimally. We further extend our schemes for the on-line case where users may join and leave. Our simulations demonstrate that the proposed algorithms achieve close to optimal load balancing and max-min fairness and they outperform commonly used heuristic approaches.

In third generation (3G) wireless data networks, repeated requests for popular data items can exacerbate the already scarce wireless spectrum. In this paper we propose an architectural and protocol framework that allows 3G service providers to host efficient content distribution services. We offload the spectrum intensive task of content distribution to an ad-hoc network. Less mobile users (resident subscribers) are provided incentives to cache popular data items while mobile users (transit subscribers) access this data from resident subscribers through the ad-hoc network. Since the participants of this data distribution network act as selfish agents, they may collude to maximize their individual payoff. Our proposed protocol discourages potential collusion scenarios. In this architecture the goal (social function) of the 3G service provider is to have the selfishly motivated resident subscribers service as many data requests as possible. However, the choice of which set of items to cache is left to the individual user. The caching activity among the different users can be modeled as a market sharing game. In this work, we study the Nash equilibria of market sharing games and the performance of such equilibria in terms of a social function. These games are a special case of congestion games that have been studied in the economics literature. In particular, pure strategy Nash equilibria for this set of games exist. We give a polynomial-time algorithm to find a pure strategy Nash equilibrium for a special case while it is is NP-Hard to do so in the general case. As for the performance of Nash equilibria, we show that the price of anarchy -- the worst-case ratio between the social function at any Nash equilibrium and at the social optimum -- can be upper bounded by a factor of 2. When the popularity follows a Zipf distribution, the price of anarchy is bounded by 1.45 in the special case where caching any item has a positive reward for all players. We prove that the selfish behavior of computationally bounded agents converges to an approximate Nash equilibrium in a finite number of improvements. Furthermore, we show that even with one improvement by each player, an O(log n) approximate solution can be obtained. Our simulation scenarios show that the price of anarchy is 30% better than that of the worst-case analysis and that the system quickly (1 or 2 steps) converges to a Nash equilibrium.

Each access point (AP) in a WiFi network must be assigned a channel for it to service users. There are only finitely many possible channels that can be assigned. Moreover, neighboring access points must use different channels so as to avoid interference. Currently these channels are assigned by administrators who carefully consider channel conflicts and network loads. Channel conflicts among APs operated by different entities are currently resolved in an ad hoc manner or not resolved at all. We view the channel assignment problem as a game, where the players are the service providers and APs are acquired sequentially. We consider the price of anarchy of this game, which is the ratio between the total coverage of the APs in the worst Nash equilibrium of the game and what the total coverage of the APs would be if the channel assignment were done by a central authority. We provide bounds on the price of anarchy depending on assumptions on the underlying network and the type of bargaining allowed between service providers. The key tool in the analysis is the identification of the Nash equilibria with the solutions to a maximal coloring problem in an appropriate graph. We relate the price of anarchy of these games to the approximation factor of local optimization algorithms for the maximum k-colorable subgraph problem. We also study the speed of convergence in these games.

The ability to locate network bottlenecks along end-to-end paths on the Internet is of great interest to both network operators and researchers. For example, knowing where bottleneck links are, network operators can apply traffic engineering either at the interdomain or intradomain level to improve routing. Existing tools either fail to identify the location of bottlenecks, or generate a large amount of probing packets. In addition, they often require access to both end points. In this paper we present Pathneck, a tool that allows end users to efficiently and accurately locate the bottleneck link on an Internet path. Pathneck is based on a novel probing technique called Recursive Packet Train (RPT) and does not require access to the destination. We evaluate Pathneck using wide area Internet experiments and trace-driven emulation. In addition, we present the results of an extensive study on bottlenecks in the Internet using carefully selected, geographically diverse probing sources and destinations. We found that Pathneck can successfully detect bottlenecks for almost 80% of the Internet paths we probed. We also report our success in using the bottleneck location and bandwidth bounds provided by Pathneck to infer bottlenecks and to avoid bottlenecks in multihoming and overlay routing.

Tracing attack packets to their sources, known as IP traceback, is an important step to counter distributed denial-of-service (DDoS) attacks. In this paper, we propose a novel packet logging based (i.e., hash-based) traceback scheme that requires an order of magnitude smaller processing and storage cost than the hash-based scheme proposed by Snoeren, et al. (2001), thereby being able to scalable to much higher link speed (e.g., OC-768). The baseline idea of our approach is to sample and log a small percentage (e.g., 3.3%) of packets. The challenge of this low sampling rate is that much more sophisticated techniques need to be used for traceback. Our solution is to construct the attack tree using the correlation between the attack packets sampled by neighboring routers. The scheme using naive independent random sampling does not perform well due to the low correlation between the packets sampled by neighboring routers. We invent a sampling scheme that improves this correlation and the overall efficiency significantly. Another major contribution of this work is that we introduce a novel information-theoretic framework for our traceback scheme to answer important questions on system parameter tuning and the fundamental trade-off between the resource used for traceback and the traceback accuracy. Simulation results based on real-world network topologies (e.g. Skitter) match very well with results from the information-theoretic analysis. The simulation results also demonstrate that our traceback scheme can achieve high accuracy, and scale very well to a large number of attackers (e.g., 5000+).

The topology of a wireless multihop network can be controlled by varying the transmission power at each node. It is not energy efficient to use the communication network G/sub max/ where every node transmits with maximum power. For energy efficient operations, it is desirable to have a subnetwork that preserves a minimum-energy path between every pair of nodes (where a minimum-energy path is one that allows messages to be transmitted with a minimum use of energy). We first identify conditions that are necessary and sufficient for a subnetwork G of G/sub max/ to preserve this property. Using this characterization, we then propose an efficient topology-control algorithm that, given a communication network G/sub max/, computes a subnetwork G that it preserves at least one minimum-energy path between every pair of nodes. We also propose an energy-efficient reconfiguration protocol that maintains this minimum-energy path property as the network topology changes dynamically. We demonstrate the performance improvements of our algorithm over other existing topology-control algorithms through simulation.

Per-flow traffic measurement is critical for usage accounting, traffic engineering, and anomaly detection. Previous methodologies are either based on random sampling (e.g., Cisco's NetFlow), which is inaccurate, or only account for the "elephants". Our paper introduces a novel technique for measuring per-flow traffic approximately, for all flows regardless of their sizes, at very high-speed (say, OC192+). The core of this technique is a novel data structure called Space Code Bloom Filter (SCBF). A SCBF is an approximate representation of a multiset; each element in this multiset is a traffic flow and its multiplicity is the number of packets in the flow. SCBF employs a Maximum Likelihood Estimation (MLE) method to measure the multiplicity of an element in the multiset. Through parameter tuning, SCBF allows for graceful tradeoff between measurement accuracy and computational and storage complexity. SCBF also contributes to the foundation of data streaming by introducing a new paradigm called blind streaming. We evaluated the performance of SCBF on packet traces gathered from a tier-1 ISP backbone and through mathematical analysis. Our preliminary results demonstrate that SCBF achieves reasonable mea-surement accuracy with very low storage and computational complexity.

Designing optimal measurement infrastructure is a key step for network management. In this work we address the problem of optimizing a scalable distributed polling system. The goal of the optimization is to reduce the cost of deployment of the measurement infrastructure by identifying a minimum poller set subject to bandwidth constraints on the individual links. We show that this problem is NP-hard and propose three different heuristics to obtain a solution. We evaluate our heuristics on both hierarchical and flat topologies with different network sizes under different polling bandwidth constraints. We find that the heuristic of choosing the poller that can poll the maximum number of unpolled nodes is the best approach. Our simulation studies show that the results obtained by our best heuristic is close to the lower bound obtained using LP relaxation.

In third-generation (3G) wireless data networks, mobile users experiencing poor channel quality usually have low data-rate connections with the base-station. Providing service to low data-rate users is required for maintaining fairness, but at the cost of reducing the cell's aggregate throughput. In this paper, we propose the Unified Cellular and Ad-Hoc Network (UCAN) architecture for enhancing cell throughput, while maintaining fairness. In UCAN, a mobile client has both 3G cellular link and IEEE 802.11-based peer-to-peer links. The 3G base station forwards packets for destination clients with poor channel quality to proxy clients with better channel quality. The proxy clients then use an ad-hoc network composed of other mobile clients and IEEE 802.11 wireless links to forward the packets to the appropriate destinations, thereby improving cell throughput. We refine the 3G base station scheduling algorithm so that the throughput gains of active clients are distributed proportional to their average channel rate, thereby maintaining fairness. With the UCAN architecture in place, we propose novel greedy and on-demand protocols for proxy discovery and ad-hoc routing that explicitly leverage the existence of the 3G infrastructure to reduce complexity and improve reliability. We further propose a secure crediting mechanism to motivate users to participate in relaying packets for others. Through extensive simulations with HDR and IEEE 802.11b, we show that the UCAN architecture can improve individual user's throughput by up to 310% and the aggregate throughput of the HDR downlink by up to 60%.

The emerging multi-protocol label switching (MPLS) networks enable network service providers to route bandwidth guaranteed paths between customer sites (see Davie, B. and Rekhter, Y., 2000; Awduche, D. et. al., 1999; Sharma, V. et al., 2002; Jamoussi et al., 2002). This basic label switched path (LSP) routing is often enhanced using restoration routing which sets up alternate LSPs to guarantee uninterrupted connectivity in case network links or nodes along the primary path fail. We address the problem of distributed routing of restoration paths, defined as follows: given a request for a bandwidth guaranteed LSP between two nodes, find a primary LSP and a set of backup LSPs that protect the links along the primary LSP. A routing algorithm that computes these paths must optimize the restoration latency and the amount of bandwidth used. We introduce the concept of "backtracking" to bound the restoration latency. We consider three different cases characterized by a parameter called backtracking distance, D: (1) no backtracking (D=0); (2) limited backtracking (D=k); (3) unlimited backtracking (D=/spl infin/). We use a link cost model that captures bandwidth sharing among links using various types of aggregate link state information. We first show that joint optimization of primary and backup paths is NP-hard in all cases. We then consider algorithms that compute primary and backup paths in two separate steps. Using link cost metrics that capture bandwidth sharing, we devise heuristics for each case. Our simulation study shows that these algorithms offer a way to tradeoff bandwidth to meet a range of restoration latency requirements.

Many ad hoc routing protocols are based on some variant of flooding. Despite various optimizations, many routing messages are propagated unnecessarily. We propose a gossiping-based approach, where each node forwards a message with some probability, to reduce the overhead of the routing protocols. Gossiping exhibits bimodal behavior in sufficiently large networks: in some executions, the gossip dies out quickly and hardly any node gets the message; in the remaining executions, a substantial fraction of the nodes gets the message. The fraction of executions in which most nodes get the message depends on the gossiping probability and the topology of the network. In the networks we have considered, using gossiping probability between 0.6 and 0.8 suffices to ensure that almost every node gets the message in almost every execution. For large networks, this simple gossiping protocol uses up to 35% fewer messages than flooding, with improved performance. Gossiping can also be combined with various optimizations of flooding to yield further benefits. Simulations show that adding gossiping to AODV results in significant performance improvement, even in networks as small as 150 nodes. We expect that the improvement should be even more significant in larger networks.

We propose a protocol that, given a communication network, computes a subnetwork such that, for every pair (u, /spl upsi/) of nodes connected in the original network, there is a a minimum-energy path between u and /spl upsi/ in the subnetwork (where a minimum-energy path is one that allows messages to be transmitted with a minimum use of energy). The network computed by our protocol is in general a subnetwork of the one computed by the protocol given by Rodoplu and Meng (see IEEE J. Selected Areas in Communications, vol.17, no.8, p.1333-44, 1999). Moreover, our protocol is computationally simpler. We demonstrate the performance improvements obtained by using the subnetwork computed by our protocol through simulation.

The topology of wireless multihop ad hoc networks can be controlled by varying the transmission power of each node. We propose a simple distributed algorithm where each node makes local decisions about its transmission power and these local decisions collectively guarantee global connectivity. Specifically, based on the directional information, a node grows it transmission power until it finds a neighbor node in every direction. The resulting network topology increases the network lifetime by reducing the transmission power and reduces traffic interference by having low node degrees. Moreover, we show that the routes in the multihop network are efficient in power consumption. We give an approximation scheme in which the power consumption of each route can be made arbitrarily close to the optimal by carefully choosing the parameters. Simulation results demonstrate significant performance improvements.

In wireless networks, mobile hosts must update the network with their current location in order to get packets delivered. Paging facilitates efficient power management at the mobile host by allowing the host to update the networkless frequently at the cost of providing the network with only approximate location information. The network determines the exact location of a mobile host through paging before delivering packets destined to the mobile host. In this paper, we propose the concept of paging as an IP service. IP paging enables a common infrastructure and protocol to support the different wireless interfaces such as CDMA, GPRS, wireless LAN, avoiding the duplication of several application layer paging implementations and the inter-operability issues that exists today. We present the design, implementation, and detailed qualitative and quantitative evaluation, using measurements and simulation, of three IP-based paging protocols for mobile hosts.

The topology of a wireless multi-hop network can be controlled by varying the transmission power at each node. In this paper, we give a detailed analysis of a cone-based distributed topology control algorithm. This algorithm, introduced in [16], does not assume that nodes have GPS information available; rather it depends only on directional information. Roughly speaking, the basic idea of the algorithm is that a node u transmits with the minimum power pu, α required to ensure that in every cone of degree α around u, there is some node that u can reach with power pu, α. We show that taking α = 5π/6 is a necessary and sufficient condition to guarantee that network connectivity is preserved. More precisely, if there is a path from s to t when every node communicates at maximum power then, if α ⪇ 5π/6, there is still a path in the smallest symmetric graph Gα containing all edges (u, v) such that u can communicate with v using power pu, α. On the other hand, if α > 5π/6, connectivity is not necessarily preserved. We also propose a set of optimizations that further reduce power consumption and prove that they retain network connectivity. Dynamic reconfiguration in the presence of failures and mobility is also discussed. Simulation results are presented to demonstrate the effectiveness of the algorithm and the optimizations.