Membership inference attacks (MIAs) pose a significant threat to the privacy of machine learning models and are widely used as tools for privacy assessment, auditing, and machine unlearning. While prior MIA research has primarily focused on performance metrics such as AUC, accuracy, and TPR@low FPR—either by developing new methods to enhance these metrics or using them to evaluate privacy solutions—we found that it overlooks the disparities among different attacks. These disparities, both between distinct attack methods and between multiple instantiations of the same method, have crucial implications for the reliability and completeness of MIAs as privacy evaluation tools. In this paper, we systematically investigate these disparities through a novel framework based on coverage and stability analysis. Extensive experiments reveal significant disparities in MIAs, their potential causes, and their broader implications for privacy evaluation. To address these challenges, we propose an ensemble framework with three distinct strategies to harness the strengths of state-of-the-art MIAs while accounting for their disparities. This framework not only enables the construction of more powerful attacks but also provides a more robust and comprehensive methodology for privacy evaluation.

With the growing adoption of privacy-preserving machine learning algorithms, such as Differentially Private Stochastic Gradient Descent (DP-SGD), training or fine-tuning models on private datasets has become increasingly prevalent. This shift has led to the need for models offering varying privacy guarantees and utility levels to satisfy diverse user requirements. Managing numerous versions of large models introduces significant operational challenges, including increased inference latency, higher resource consumption, and elevated costs. Model deduplication is a technique widely used by many model serving and database systems to support high-performance and low-cost inference queries and model diagnosis queries. However, none of the existing model deduplication works has considered privacy, leading to unbounded aggregation of privacy costs for certain deduplicated models and inefficiencies when applied to deduplicate DP-trained models. We formalize the problem of deduplicating DP-trained models for the first time and propose a novel privacy- and accuracy-aware deduplication mechanism to address the problem. We developed a greedy strategy to select and assign base models to target models to minimize storage and privacy costs. When deduplicating a target model, we dynamically schedule accuracy validations and apply the Sparse Vector Technique to reduce the privacy costs associated with private validation data. Compared to baselines, our approach improved the compression ratio by up to 35× for individual models (including large language models and vision transformers). We also observed up to 43× inference speedup due to the reduction of I/O operations.

Decision forest, including RandomForest, XGBoost, and LightGBM, dominates the machine learning tasks over tabular data. Recently, several frameworks were developed for decision forest inference, such as ONNX, TreeLite from Amazon, TensorFlow Decision Forest from Google, HummingBird from Microsoft, Nvidia FIL, and lleaves. While these frameworks are fully optimized for inference computations, they are all decoupled with databases and general data management frameworks, which leads to cross-system performance overheads. We first provided a DICT model to understand the performance gaps between decoupled and in-database inference. We further identified that for in-database inference, in addition to the popular UDF-centric representation that encapsulates the ML into one User Defined Function (UDF), there also exists a relation-centric representation that breaks down the decision forest inference into several fine-grained SQL operations. The relation-centric representation can achieve significantly better performance for large models. We optimized both implementations and conducted a comprehensive benchmark to compare these two implementations to the aforementioned decoupled inference pipelines and existing in-database inference pipelines such as Spark-SQL and PostgresML. The evaluation results validated the DICT model and demonstrated the superior performance of our in-database inference design compared to the baselines.

Protecting sensitive information in diagnostic data such as logs, is a critical concern in the industrial software diagnosis and debugging process. While there are many tools developed to automatically redact the logs for identifying and removing sensitive information, they have severe limitations which can cause either over redaction and loss of critical diagnostic information (false positives), or disclosure of sensitive information (false negatives), or both. To address the problem, in this paper, we argue for a source code analysis approach for log redaction. To identify a log message containing sensitive information, our method locates the corresponding log statement in the source code with logger code augmentation, and checks if the log statement outputs data from sensitive sources by using the data flow graph built from the source code. Appropriate redaction rules are further applied depending on the sensitiveness of the data sources to preserve the privacy information in the logs. We conducted experimental evaluation and comparison with other popular baselines. The results demonstrate that our approach can significantly improve the detection precision of the sensitive information and reduce both false positives and negatives.

Serving deep learning models from relational databases brings significant benefits. First, features extracted from databases do not need to be transferred to any decoupled deep learning systems for inferences, and thus the system management overhead can be significantly reduced. Second, in a relational database, data management along the storage hierarchy is fully integrated with query processing, and thus it can continue model serving even if the working set size exceeds the available memory. Applying model deduplication can greatly reduce the storage space, memory footprint, cache misses, and inference latency. However, existing data deduplication techniques are not applicable to the deep learning model serving applications in relational databases. They do not consider the impacts on model inference accuracy as well as the inconsistency between tensor blocks and database pages. This work proposed synergistic storage optimization techniques for duplication detection, page packing, and caching, to enhance database systems for model serving. Evaluation results show that our proposed techniques significantly improved the storage efficiency and the model inference latency, and outperformed existing deep learning frameworks in targeting scenarios.

More and more organizations move their data and workload to commercial cloud storage systems. However, the multiplexing and sharing of the resources in a cloud storage system present unpredictable data access latency to tenants, which may make online data-intensive applications unable to satisfy their deadline requirements. Thus, it is important for cloud storage systems to provide deadline guaranteed services. In this paper, to meet a current form of service level objective (SLO) that constrains the percentage of each tenant's data access requests failing to meet its required deadline below a given threshold, we build a mathematical model to derive the upper bound of acceptable request arrival rate on each server. We then propose a Deadline Guaranteed storage service (called DGCloud) that incorporates three basic algorithms. Its deadline-aware load balancing scheme redirects requests and creates replicas to release the excess load of each server beyond the derived upper bound. Its workload consolidation algorithm tries to maximally reduce servers while still satisfying the SLO to maximize the resource utilization. Its data placement optimization algorithm re-schedules the data placement to minimize the transmission cost of data replication. We further propose three enhancement methods to further improve the performance of DGCloud. A dynamic load balancing method allows an overloaded server to quickly offload its excess workload. A data request queue improvement method sets different priorities to the data responses in a server's queue so that more requests can satisfy the SLO requirement. A wakeup server selection method selects a sleeping server that stores more popular data to wake up, which allows it to handle more data requests. Our trace-driven experiments in simulation and Amazon EC2 show the superior performance of DGCloud compared with previous methods in terms of deadline guarantees and system resource utilization, and the effectiveness of its individual algorithms.

Membership inference attacks seek to infer membership of individual training instances of a model to which an adversary has black-box access through a machine learning-as-a-service API. In providing an in-depth characterization of membership privacy risks against machine learning models, this paper presents a comprehensive study towards demystifying membership inference attacks from two complimentary perspectives. First, we provide a generalized formulation of the development of a black-box membership inference attack model. Second, we characterize the importance of model choice on model vulnerability through a systematic evaluation of a variety of machine learning models and model combinations using multiple datasets. Through formal analysis and empirical evidence from extensive experimentation, we characterize under what conditions a model may be vulnerable to such black-box membership inference attacks. We show that membership inference vulnerability is data-driven and corresponding attack models are largely transferable. Though different model types display different vulnerabilities to membership inference, so do different datasets. Our empirical results additionally show that (1) using the type of target model under attack within the attack model may not increase attack effectiveness and (2) collaborative learning exposes vulnerabilities to membership inference risks when the adversary is a participant. We also discuss countermeasure and mitigation strategies.

Cloud computing offers a cost-effective and elastic computing paradigm that facilitates large scale data storage and analytics. By deploying virtualization technologies in the datacenter, cloud enables efficient resource management and isolation for various big data applications. Since the hotspots (i.e., overloaded machines) can degrade the performance of these applications, virtual machine migration has been utilized to perform load balancing in the datacenters to eliminate hotspots and guarantee Service Level Agreements (SLAs). However, the previous load balancing schemes make migration decisions based on deterministic resource demand estimation and workload characterization, without considering their stochastic properties. By studying real world traces, we show that the resource demand and workload of virtual machines are highly dynamic and bursty, which can cause these schemes to make inefficient migrations for load balancing. To address this problem, in this paper we propose a stochastic load balancing scheme which aims to provide probabilistic guarantee against the resource overloading with virtual machine migration, while minimizing the total migration overhead. Our scheme effectively addresses the prediction of the distribution of resource demand and the multidimensional resource requirements with stochastic characterization. Moreover, as opposed to the previous works that measure the migration cost without considering the network topology, our scheme explicitly takes into account the distance between the source physical machine and the destination physical machine for a virtual machine migration. The trace-driven experiments show that our scheme outperforms the previous schemes in terms of SLA violation and the migration cost.

Learning Rate (LR) is an important hyper-parameter to tune for effective training of deep neural networks (DNNs). Even for the baseline of a constant learning rate, it is non-trivial to choose a good constant value for training a DNN. Dynamic learning rates involve multi-step tuning of LR values at various stages of the training process and offer high accuracy and fast convergence. However, they are much harder to tune. In this paper, we present a comprehensive study of 13 learning rate functions and their associated LR policies by examining their range parameters, step parameters, and value update parameters. We propose a set of metrics for evaluating and selecting LR policies, including the classification confidence, variance, cost, and robustness, and implement them in LRBench, an LR benchmarking system. LRBench can assist end-users and DNN developers to select good LR policies and avoid bad LR policies for training their DNNs. We tested LRBench on Caffe, an open source deep learning framework, to showcase the tuning optimization of LR policies. Evaluated through extensive experiments, we attempt to demystify the tuning of LR policies by identifying good LR policies with effective LR value ranges and step sizes for LR update schedules.

Deep learning techniques based on neural networks have shown significant success in a wide range of AI tasks. Large-scale training datasets are one of the critical factors for their success. However, when the training datasets are crowdsourced from individuals and contain sensitive information, the model parameters may encode private information and bear the risks of privacy leakage. The recent growing trend of the sharing and publishing of pre-trained models further aggravates such privacy risks. To tackle this problem, we propose a differentially private approach for training neural networks. Our approach includes several new techniques for optimizing both privacy loss and model accuracy. We employ a generalization of differential privacy called concentrated differential privacy(CDP), with both a formal and refined privacy loss analysis on two different data batching methods. We implement a dynamic privacy budget allocator over the course of training to improve model accuracy. Extensive experiments demonstrate that our approach effectively improves privacy loss accounting, training efficiency and model quality under a given privacy budget.

The universal popularity of GPS-enabled mobile devices and traffic navigation services has fueled the growth of trajectory data, as evidenced by Uber Movement and NYC taxi data release. Although trajectory data can generate valuable insights and value-added services for many, publishing this data while respecting mobile users' privacy has been a long-standing challenge. In this paper, we present DP-Star, a methodical framework for publishing trajectory data with differential privacy guarantee as well as high utility preservation. DP-Star relies on a novel combination of several components. First, DP-Star's normalization algorithm uses the Minimum Description Length metric to summarize raw trajectories using their representative points, thereby achieving a desirable trade-off between the preciseness and conciseness of their information content. Second, DP-Star constructs a density-aware grid which ensures spatial densities can be preserved despite the noise added to satisfy differential privacy. Third, DP-Star preserves the correlations between trajectories' end points through a private trip distribution, and intermediate points through a private Markov mobility model. Finally, DP-Star estimates users' trip lengths using a median length estimation method, and generates synthetic trajectories that preserve both differential privacy and high utility. Our experimental comparison shows that DP-Star significantly outperforms existing approaches in terms of trajectory utility and accuracy.

Data collection is a fundamental and critical function of wireless sensor networks (WSNs) for the cyber-physical systems (CPS) to estimate the state of the physical world. However, unstable network conditions impose significant challenges in guaranteeing the data accuracy that is essential for the reliable estimation of physical states. Without efficiently resolving congestion during data transmission in WSNs, packet loss due to congestion can significantly degrade the data quality. Various congestion control schemes have been proposed to address this issue. Most of them rely on reducing transmitted data samples to eliminate the congestion, which, however, could lead to abysmally high estimation error. In this paper, we analyze the impact of congestion control on the data accuracy and propose a Congestion-Adaptive Data Collection scheme (CADC) to efficiently resolve the congestion under the guarantee of data accuracy. CADC mitigates congestion by adaptive lossy compression while ensuring a given overall data estimation error bound in a distributed manner. Considering that for a CPS application different data items may have different priorities, we also propose a weighted CADC scheme such that the data with higher priority has less distortion. We further adapt CADC to guarantee the accuracy of specific aggregate computations. Extensive simulations demonstrate the effectiveness and efficiency of CADC.

As mobile devices and location-based services become increasingly ubiquitous, the privacy of mobile users' location traces continues to be a major concern. Traditional privacy solutions rely on perturbing each position in a user's trace and replacing it with a fake location. However, recent studies have shown that such point-based perturbation of locations is susceptible to inference attacks and suffers from serious utility losses, because it disregards the moving trajectory and continuity in full location traces. In this paper, we argue that privacy-preserving synthesis of complete location traces can be an effective solution to this problem. We present AdaTrace, a scalable location trace synthesizer with three novel features: provable statistical privacy, deterministic attack resilience, and strong utility preservation. AdaTrace builds a generative model from a given set of real traces through a four-phase synthesis process consisting of feature extraction, synopsis learning, privacy and utility preserving noise injection, and generation of differentially private synthetic location traces. The output traces crafted by AdaTrace preserve utility-critical information existing in real traces, and are robust against known location trace attacks. We validate the effectiveness of AdaTrace by comparing it with three state of the art approaches (ngram, DPT, and SGLT) using real location trace datasets (Geolife and Taxi) as well as a simulated dataset of 50,000 vehicles in Oldenburg, Germany. AdaTrace offers up to 3-fold improvement in trajectory utility, and is orders of magnitude faster than previous work, while preserving differential privacy and attack resilience.

The exceptional increase in the usage of smartphones has contributed to a massive increase in data traffic from application servers to the smartphones, which not only strains their computation capacities and batteries but also bogs down the last hop in data transmission. For this problem, traffic redundancy elimination (TRE) is an effective solution, in which a chunk to be transmitted could be directly fetched from the receiver's cache. However, existing TRE solutions either cannot be directly applied to or are not suitable for smartphones due to high computing and energy overhead imposed on smartphones. To address this problem, in this paper, we propose a novel TRE system, called TailoredRE, which consists of three components. First, each smartphone has a clone in the cloud that is responsible for computation intensive tasks including parsing traffic and detecting redundancy. Second, considering that each mobile user has certain applications (e.g., YouTube) to use in daily life, each smartphone's clone selectively chooses the applications that are most frequently used by the user and also have high redundancy ratios to cache data. Third, considering that some users always have common favorite applications, TailoredRE clusters their clones together to cooperatively conduct the redundancy detection task in order to reduce the cache resource consumption in the cloud. We collected traces from eleven applications including Web Browser, YouTube, CNN, Quora, Instagram and Facebook, and used the traces in simulation. We also implemented and open-sourced TailoredRE and conducted prototype-based experiments. Experiment results show that TailoredRE can achieve much higher cache hit rate, end-to-end throughput, bandwidth saving and energy efficiency compared with previous TRE methods.

As a popular routing protocol in wireless sensor networks (WSNs), greedy routing has received great attention. The previous works characterize its data deliverability in WSNs by the probability of all nodes successfully sending their data to the base station. Their analysis, however, neither provides the information of the quantitative relation between successful data delivery ratio and transmission power of sensor nodes nor considers the impact of the network congestion or link collision on the data deliverability. To address these problems, in this paper, we characterize the data deliverability of greedy routing by the ratio of successful data transmissions from sensors to the base station. We introduce n-guaranteed delivery which means that the ratio of successful data deliveries is not less than n, and study the relationship between the transmission power of sensors and the probability of achieving n-guaranteed delivery. Furthermore, with considering the effect of network congestion, link collision, and holes (e.g., those caused by physical obstacles such as a lake), we provide a more precise and full characterization for the deliverability of greedy routing. Extensive simulation and real-world experimental results show the correctness and tightness of the upper bound of the smallest transmission power for achieving n-guaranteed delivery.

In the cloud, multiple tenants share the resource of datacenters and their applications compete with each other for scarce network bandwidth. Current studies have shown that the lack of bandwidth guarantee causes unpredictable network performance, leading to poor application performance. To address this issue, several virtual network abstractions have been proposed which allow the tenants to reserve virtual clusters with specified bandwidth between the Virtual Machines (VMs) in the datacenters. However, all these existing proposals require the tenants to deterministically characterize the bandwidth demands in the abstractions, which can be difficult and result in inefficient bandwidth reservation due to the demand uncertainty. In this paper, we explore a virtual cluster abstraction with stochastic bandwidth characterization to address the bandwidth demand uncertainty. We propose Stochastic Virtual Cluster (SVC), which models the bandwidth demand between VMs in a probabilistic way. Based on SVC, we develop a stochastic framework for virtual cluster allocation, in which the admitted virtual cluster's bandwidth demands are satisfied with a high probability. Efficient VM allocation algorithms are proposed to implement the framework while reducing the possibility of link congestion through minimizing the maximum bandwidth occupancy of a virtual cluster on physical links. Using simulations, we show that SVC achieves the trade-off between the job concurrency and the average job running time, and demonstrate its effectiveness for accommodating cloud application workloads with highly volatile bandwidth demands and its improvement to work-conserving bandwidth enforcement.

This paper aims to improve wireless network throughput by suppressing duplicate data transmissions from network links. It has been demonstrated that IP-layer Redundancy Elimination (RE) with content overhearing can significantly improve the goodput and utilization of wireless channels in wireless environment. However, the integration of IP-layer RE and wireless overhearing introduces a challenge. That is, probabilistic wireless overhearing and the possibility of a receiver overhearing from multiple transmitters cause the caches of a sender and a receiver far from synchronization, which can disrupt IP-layer RE's correctness and degrade its performance. The previous work deals with this challenge by the overhearing probability estimation, which however is not efficient or scalable. In this paper, we propose a Prediction-based Redundancy Elimination with Content Overhearing method (PRECO) to address this challenge. By exploiting prediction-based RE, PRECO does not require cache synchronization and overhearing probability estimation, which enables its efficient and scalable deployment. Based on PRECO, we exploit the benefits of deploying sub-packet level RE as a primitive IP-layer service on all nodes in wireless mesh networks by proposing a redundancy-aware routing protocol. Trace-driven performance evaluation shows the effectiveness and efficiency of PRECO compared with other RE methods.

The pay-as-you-go service model impels cloud customers to reduce the usage cost of bandwidth. Traffic Redundancy Elimination (TRE) has been shown to be an effective solution for reducing bandwidth costs, and thus has recently captured significant attention in the cloud environment. By studying the TRE techniques in a trace driven approach, we found that both short-term (time span of seconds) and long-term (time span of hours or days) data redundancy can concurrently appear in the traffic, and solely using either sender-based TRE or receiver-based TRE cannot simultaneously capture both types of traffic redundancy. Also, the efficiency of existing receiver-based TRE solution is susceptible to the data changes compared to the historical data in the cache. In this paper, we propose a Cooperative end-to-end TRE solution (CoRE) that can detect and remove both short-term and long-term redundancy through a two-layer TRE design with cooperative operations between layers. An adaptive prediction algorithm is further proposed to improve TRE efficiency through dynamically adjusting the prediction window size based on the hit ratio of historical predictions. Besides, we enhance CoRE to adapt to different traffic redundancy characteristics of cloud applications to improve its operation cost. Extensive evaluation with several real traces show that CoRE is capable of effectively identifying both short-term and long-term redundancy with low additional cost while ensuring TRE efficiency from data changes.

Community question answering services (CQAS) (e.g., Yahoo! Answers) provides a platform where people post questions and answer questions posed by others. Previous works analyzed the answer quality (AQ) based on answer-related features, but neglect the question-related features on AQ. Previous work analyzed how asker- and question-related features affect the question quality (QQ) regarding the amount of attention from users, the number of answers and the question solving latency, but neglect the correlation between QQ and AQ (measured by the rating of the best answer), which is critical to quality of service (QoS). We handle this problem from two aspects. First, we additionally use QQ in measuring AQ, and analyze the correlation between a comprehensive list of features (including answer-related features) and QQ. Second, we propose the first method that estimates the probability for a given question to obtain high AQ. Our analysis on the Yahoo! Answers trace confirmed that the list of our identified features exert influence on AQ, which determines QQ. For the correlation analysis, the previous classification algorithms cannot consider the mutual interactions between multiple (>2) classes of features. We then propose a novel Coupled Semi-Supervised Mutual Reinforcement-based Label Propagation (CSMRLP) algorithm for this purpose. Our extensive experiments show that CSMRLP outperforms the Mutual Reinforcement-based Label Propagation (MRLP) and five other traditional classification algorithms in the accuracy of AQ classification, and the effectiveness of our proposed method in AQ prediction. Finally, we provide suggestions on how to create a question that will receive high AQ, which can be exploited to improve the QoS of CQAS.

More and more organizations move their data and workload to commercial cloud storage systems. However, the multiplexing and sharing of the resources in a cloud storage system present unpredictable data access latency to tenants, which may make online data-intensive applications unable to satisfy their deadline requirements. Thus, it is important for cloud storage systems to provide deadline guaranteed services. In this paper, to meet a current form of service level objective (SLO) that constrains the percentage of each tenant's data access requests failing to meet its required deadline below a given threshold, we build a mathematical model to derive the upper bound of acceptable request arrival rate on each server. We then propose a Deadline Guaranteed storage service (called DGCloud) that incorporates three algorithms. Its deadline-aware load balancing scheme redirects requests and creates replicas to release the excess load of each server beyond the derived upper bound. Its workload consolidation algorithm tries to maximally reduce servers while still satisfying the SLO to maximize the resource utilization. Its data placement optimization algorithm re-schedules the data placement to minimize the transmission cost of data replication. Our trace-driven experiments in simulation and Amazon EC2 show the higher performance of DGCloud compared with previous methods in terms of deadline guarantees and system resource utilization, and the effectiveness of its individual algorithms.

The shared nature of cloud network infrastructures causes unpredictable network performance, which may degrade the performance of these applications. Recently, several works propose to explicitly reserve the network bandwidth in the cloud with virtual network abstraction models, which pre-specify the network bandwidth between virtual machines (VMs) for a tenant job. However, the pre-specification fails to exploit the elastic feature of the bandwidth resource (i.e., more reserved bandwidth within no-elongation threshold bandwidth leads to shorter job execution time and vice versa) in job scheduling. It is difficult for ordinary tenants (without specialized network knowledge) to estimate the exact needed bandwidth. In this paper, we propose a new cloud job scheduler, in which each tenant only needs to specify job deadline and each job's reserved bandwidth is elastically determined by leveraging the elastic feature to maximize the total job rewards, which represent the worth of successful completion by deadlines. Finally, the scheduler tries to reduce the execution time of each job. It also jointly considers the computational capacity of VMs and reserved VM bandwidth in job scheduling. Using trace-driven and real cluster experiments, we show the efficiency and effectiveness of our job scheduler in comparison with other scheduling strategies.

Maximizing data locality in task scheduling is critical for the performance of MapReduce job execution. Manyexisting works on MapReduce scheduling decide the placementof map and reduce tasks on a coarse granularity of locationsmeasured by located machines and racks. They do not explicitlyconsider the network topology and data transmission cost, whichmay cause task straggling and degrade the job performance. Inorder to improve MapReduce job performance, in this paper, we consider the task placement with the goal of minimizing theoverall data transmission cost for a job execution while balancingthe transmission cost reduction and resource utilization. Wepropose a probabilistic network-aware scheduling algorithm thatselects a task (map task or reduce task) to be scheduled on a givenavailable task slot that leads to the minimum transmission costamong the task candidates, and then schedule the selected taskon the slot with a probability determined by its transmission cost, a lower expected transmission cost leads to a higher probabilityand vice versa. We also propose a method to more accuratelyestimate the intermediate data size based on the progress ofmap tasks, which is needed to calculate the transmission cost ofreduce tasks but is unknown at the time of reduce task scheduling. We implement our probabilistic network-aware schedulingalgorithm on Apache Hadoop and conduct experiments on ahigh-performance computing platform. The experimental resultsshow that our scheduling algorithm outperforms the previousapproaches in terms of job completion time and cluster resource utilization.

Network virtualization with bandwidth guarantee is essential for the performance predictability of cloud applications because of the shared multi-tenant nature of the cloud. Several virtual network abstractions have been proposed for the tenants to specify and reserve their virtual clusters with bandwidth guarantee. However, they require pre-determined fixed cluster size and bandwidth, and do not support the scaling of the cluster in size and bandwidth requirements. On the other hand, the existing works on virtual cluster scaling focus on dynamically adjusting the cluster size without considering any bandwidth guarantee targeted by current network abstractions. To fill the gap, this paper considers the problem of scaling up a virtual network abstraction with bandwidth guarantee. Efficient algorithms are proposed to find the valid allocation for the scaled cluster abstraction with optimization on the VM locality of the cluster. We also point out the case that a virtual cluster cannot be scaled without changing its original VM placement, and propose an optimal allocation algorithm that exploits the VM migration to address this issue while minimizing the total migration cost for the virtual cluster scaling. Extensive simulations demonstrate the effectiveness and efficiency of our algorithms.

Though a cooperative broadcast scheme has been proposed for fading environments, it has two defects: First, it only handles a packet flow from a single source node in the network, but does not consider the scenario of multiple packet flows simultaneously broadcasted from different source nodes. Second, it only allows a single relay node to forward a packet in each time slot, though multiple relay nodes forwarding in a time slot can significantly reduce broadcast latency. In this paper, we aim achieve low-latency multi-flow broadcast in wireless multi-hop networks with fading channels. To describe the interference among the transmission in different flows, we incorporate the Rayleigh fading model to the signal to noise ratio (SNR) model. Then, we introduce a cooperative diversity scheme which allows multiple relays forwarding in a time slot to reduce broadcast latency. We then formulate an interesting problem: In a fading environment, what is the optimal relay allocation schedule to minimize the broadcast latency? We propose a warm up heuristic algorithm for single-flow cooperative broadcast, based on which, we further propose a heuristic algorithm for multi-flow cooperative broadcast. Simulation results demonstrate that the two algorithms achieve lower broadcast latency than a previous method.

Cooperative adaptive cruise control (CACC) systems have the potential to increase traffic throughput by allowing smaller headway between vehicles and moving vehicles safely in a platoon at a harmonized speed. CACC systems have been attracting significant attention from both academia and industry since connectivity between vehicles will become mandatory for new vehicles in the USA in the near future. In this paper, we review three basic and important aspects of CACC systems: communications, driver characteristics, and controls to identify the most challenging issues for their real-world deployment. Different routing protocols that support the data communication requirements between vehicles in the CACC platoon are reviewed. Promising and suitable protocols are identified. Driver characteristics related issues, such as how to keep drivers engaged in driving tasks during CACC operations, are discussed. To achieve mass acceptance, the control design needs to depict real-world traffic variability such as communication effects, driver behavior, and traffic composition. Thus, this paper also discusses the issues that existing CACC control modules face when considering close to ideal driving conditions.

In this paper, we study the minimum energy broadcast problem in time-varying graphs (TVGs), which are a very useful high level abstraction for studying highly dynamic wireless networks. To this end, we first incorporate a channel model, called energy-demand functions, to the current TVGs, namely time-varying energy-demand graphs (TVEGs). Based on this model, we formulate the problem: given a TVEG, what is the optimal schedule (i.e., Which nodes should forward a packet in what times and at what power levels) to minimize the energy consumption of the broadcast? We prove the problem to be NP-hard and o(log N) in approximable. It is a challenge to find a solution for this problem on continuous time. Fortunately, we prove that the problem on continuous time is equivalent to the problem on certain discrete time points, called discrete time set (DTS). Based on this property, we propose polynomial time solutions for this problem with different channel models, and evaluate the performance of these methods from real-life contact traces.

Data collection by wireless sensor networks is a fundamental and critical function for cyber-physical systems (CPS) to estimate the state of the physical world. However, unstable network conditions impose great challenges in guaranteeing data accuracy, which is essential for reliable state estimation of the physical phenomena. For underlying sensor networks, without efficiently resolving congestion in data transmission, packet loss at congested nodes can considerably increase the estimation error. However, previous congestion control schemes relying on reducing transmitted data samples also increase the estimation error. Thus, we propose a Congestion-Adaptive Data Collection scheme (CADC) to efficiently resolve the network congestion while guaranteeing the overall data estimation accuracy. CADC mitigates congestion by adaptive lossy compression with guarantee that a given overall data estimation error bound is satisfied. Besides, since a CPS application may have different priorities for different data items, we further propose a weighted CADC scheme such that the data with higher priority has less distortion. Extensive experimental results demonstrate the effectiveness and efficiency of our CADC schemes.

As a popular routing protocol in wireless sensor networks (WSNs), greedy routing has received great attention. The previous works characterize its data deliverability in WSNs by the probability of all nodes successfully sending their data to the base station. Their analysis, however, neither provides the information of the quantitative relation between successful data delivery ratio and transmission power of sensor nodes nor considers the impact of the network congestion or link collision on the data deliverability. To address these problems, in this paper, we characterize the data deliverability of greedy routing by the ratio of successful data transmissions from sensors to the base station. We introduce η-guaranteed delivery which means that the ratio of successful data deliveries is not less than η, and study the relationship between the transmission power of sensors and the probability of achieving η-guaranteed delivery. Furthermore, with considering the effect of network congestion and link collision, we provide a more precise and full characterization for the deliverability of greedy routing. Extensive simulation and real-world experimental results show the correctness and tightness of the upper bound of the smallest transmission power for achieving η-guaranteed delivery.

In a hybrid wireless network that combines a mobile ad-hoc network and an infrastructure network, efficient and reliable data routing is important for high throughput. Existing routing schemes that simply combine ad-hoc and infrastructure routings inherit the drawbacks of ad-hoc routing including congestion and high overhead for route discovery and maintenance. Although current reputation systems help increase routing reliability, they rely on local information exchanges between nodes to evaluate node reputations, so they are not sufficiently effective and efficient. A challenge here is if we can coordinately develop an efficient routing algorithm and effective cooperation incentives for reliable routing. To handle this challenge, this paper presents a peer-to-peer (P2P)-based Market-guided Distributed Routing mechanism (MDR). MDR takes advantage of widespread base stations to coordinately realize highly efficient data routing, and effective reputation management and trading market management for reliable data routing. The packets from a source node are distributively transmitted to base stations directly or indirectly, and then they are transmitted to the destination. The base stations form a P2P structure for reputation collection and querying to avoid local information exchanges, and for managing the service transactions between nodes in the trading market. By leveraging the single-relay transmission feature, base stations can monitor the actual transmitted packets of relay nodes to more accurately and efficiently evaluate their reputations and execute trading market management, as well as detect falsely reported reputation information. We further propose market-based policies to strengthen cooperation incentives. Simulation results show that MDR outperforms the traditional hybrid routing schemes and reputation systems in achieving high throughput.

The shared multi-tenant nature of cloud network infrastructures has caused poor application performance in the clouds due to unpredictable network performance. To provide bandwidth guarantee, several virtual network abstractions have been proposed which allow the tenants to specify and reserve virtual clusters with required network bandwidth between the VMs. However, all of these existing proposals require the tenants to deterministically characterize the exact bandwidth demands in the abstractions, which can be difficult and result in inefficient bandwidth reservation due to the demand uncertainty. In this paper, we propose a virtual cluster abstraction with stochastic bandwidth requirements between VMs, called Stochastic Virtual Cluster (SVC), which probabilistically models the bandwidth demand uncertainty. Based on SVC, we propose a network sharing framework and efficient VM allocation algorithms to ensure that the bandwidth demands of tenants on any link are satisfied with a high probability, while minimizing the bandwidth occupancy cost on links. Using simulations, we demonstrate the effectiveness of SVC for accommodating cloud application workloads with highly volatile bandwidth demands, in the way of achieving the trade-off between the job concurrency and average job running time.

Cooperative broadcast, in which receivers are allowed to combine received packet from different senders to combat transmission errors, has gained increasing attention. Previous studies showed that broadcast optimization solutions are sufficient in non-fading environments but may suffer a low delivery ratio under wireless channel fading. Though previous work analyzed the tradeoff between energy and delay in cooperative broadcast, no works investigated the tradeoff in a fading environment. Thus, in this paper, we study this tradeoff with the consideration of fading. We formulate this problem as a Fading-resistant Delay-constrained Minimum Energy Cooperative Broadcast (FDMECB) problem, and prove that it is NP-complete. We then propose an approximation algorithm for theoretical interests. We further propose a heuristic algorithm that makes approximately optimal local decision to achieve global optimization. Our experimental results show that our algorithms outperform a previous non-fading resistant algorithm.

Radio frequency identification (RFID) and wireless sensor networks (WSNs) have been popular in the industrial field, and both have undergone dramatic development. RFID and WSNs are well known for their abilities in identity identification and data transmission, respectively, and hence widely used in applications for environmental and health monitoring. Though the integration of a sensor and an RFID tag was proposed to gather both RFID tag and sensed information, few previous research efforts explore the integration of data transmission modes in the RFID and WSN systems to enhance the performance of the applications. In this paper, we propose a hybrid RFID and WSN system (HRW) that synergistically integrates the traditional RFID system and WSN system for efficient data collection. HRW has hybrid smart nodes that combine the function of RFID tags, the reduced function of RFID readers, and wireless sensors. Therefore, nodes can read each other's sensed data in tags, and all data can be quickly transmitted to an RFID reader through the node that first reaches it. The RFID readers transmit the collected data to the back-end servers for data processing and management. We also propose methods to improve data transmission efficiency and to protect data privacy and avoid malicious data selective forwarding in data transmission. Comprehensive simulation and trace-driven experimental results show the high performance of HRW in terms of the cost of deployment, transmission delay and capability, and tag capacity requirement.

Continuous aggregation is usually required in many sensor applications to obtain the temporal variation information of aggregates. However, in a hostile environment, the adversary could fabricate false temporal variation patterns of the aggregates by manipulating a series of aggregation results through compromised nodes. Existing secure aggregation schemes conduct one individual verification for each aggregation result, which could incur great accumulative communication cost and negative impact on transmission scheduling for continuous aggregation. In this paper, we identify distinct design issues for protecting continuous in-network aggregation and propose a novel scheme to detect false temporal variation patterns. Compared with the existing schemes, our scheme greatly reduces the verification cost by checking only a small part of aggregation results to verify the correctness of the temporal variation patterns in a time window. A sampling-based approach is used to check the aggregation results, which enables our scheme independent of any particular in-network aggregation protocols as opposed to existing schemes. We also propose a series of security mechanisms to protect the sampling process. Both theoretical analysis and simulations show the effectiveness and efficiency of our scheme.

Cooperative broadcast, in which a packet receiver cooperatively combines received weak signal power from different senders to decode the original packet, has gained increasing attention. However, existing approaches are developed based on the assumption that there is a single flow in the network; thus, they are not suitable for multi-flow broadcasting in which broadcasts are initiated by different nodes and consist of more than one packet at any point in time. In this paper, we aim to achieve low-latency multi-flow broadcast in wireless multihop networks with fading channels. We formulate this problem as a Minimum Slotted Delay Cooperative Broadcast (MSDCB) problem, and prove that it is NP-complete and o(logN) inapproximable. We then propose two heuristic algorithms named PCBHS and PCBH-M to solve MSDCB. Our experimental results show that our algorithms outperform previous methods.

The pay-as-you-go service model impels cloud customers to reduce the usage cost of bandwidth. Traffic Redundancy Elimination (TRE) has been shown to be an effective solution for reducing bandwidth costs, and has recently captured significant attention in the cloud environment. By studying the TRE techniques with a trace driven approach, we found that solely using either sender-based TRE or receiver-based TRE cannot simultaneously capture traffic redundancy in both short-term (time span of seconds) and long-term (time span of hours or days) data redundancy, which concurrently appear in the traffic. Additionally, the TRE efficiency of existing receiver-based TRE solution is susceptible to data changes compared to historical data in the cache. In this paper, we propose a sender and receiver Cooperative end-to-end TRE solution (CoRE) for efficiently identifying and removing both short-term and long-term redundancy. Through a two-layer redundancy detection design and one single pass algorithm for chunking and fingerprinting, CoRE efficiently carries out cooperative operations between the sender and the receiver. By extensive evaluation with several real traces, we show that CoRE is able to identify both short-term and longterm redundancy with low additional cost, while ensuring TRE efficiency from data changes.

In the applications of wireless sensor networks, the peak values, such as largest sensed values and their locations, are very useful for detecting abnormal events happened in the monitored region. Although the results returned by the traditional top-k queries provide k largest sensed values, they ignore the spatial-correlation of the sensed data so that the locations of the returned values are very close to each other and only tell a small area being abnormal or few number of abnormal events happening. Due to this reason, the Location Aware Peak Value Query, denoted by LAP-(D,k) query, is proposed in this paper. For any given D and k, the LAP-(D,k) query returns k largest sensed values and their locations, and the distance between the any two locations is larger than D. The problem of processing LAP-(D,k) query is proved to be NP-hard, and two distributed approximation algorithms are proposed to solve this problem. One is a distributed greedy algorithm with ratio bound 5.8. The other one is a region partition based algorithm with ratio bound 3. The theoretical analysis and experimental results show that the proposed algorithms have high performance in terms of accuracy and energy consumption.

When monitoring environments with wireless sensor networks, optimal sensor deployment is a fundamental issue and an effective means to achieve desired performance. Selecting best sensor deployment has a dependence on the deployment environments. Existing works address sensor deployment within three types of environments including one dimensional line, 2-D field and 3-D space. However, in many applications the deployment environments usually have network structures, which cannot be simply classified as the three types. The deployed locations and communications of sensor nodes are limited onto the network edges, which make the deployment problem distinct from that in other types of environments. In this paper, we study sensor deployment in network-structured environments and aim to achieve k-coverage while minimizing the number of sensor nodes. Furthermore, we jointly consider the optimization of sink deployment and routing strategies with the goal to minimize the network communication cost of data collection. To the best of our knowledge, this paper is the first one to tackle sensor/sink deployment under the deployment constraints imposed by the network structure. The hardness of the problems is shown. Polynomial-time algorithms are proposed to determine optimal sensor/sink deployment and routing strategies in tree-topology network structure. Efficient approximation algorithms are proposed for the general graph network structure and their performances are analyzed. Theoretical results and extensive simulations show the efficiency of the proposed algorithms.

Wireless sensor networks(WSNs) are being widely deployed for many monitoring applications, of which a popular one is anti-theft. However, current WSN technologies for anti-theft are either very susceptible to the environment interference or easily compromised by the thieves. Hence, they fail to achieve the desired effectiveness in many anti-theft scenarios. To address these limitations, this paper proposes a novel Hidden Anti-theft System (HAS) to monitor theft intrusion, which is based on the influences of theft intrusion on signal strength according to the shadowing effect in wireless communication. The theft intrusion is detected by finding abnormal RSSI samples of a wireless link compared with the stable range of signal strength in the link's normal state. Through the proposed monitoring approach, HAS can effectively detect intrusion while keeping invisible. In HAS system, to achieve load balance for detection task, we propose an efficient algorithm to determine the set of links that each node monitors. To reduce the response time, a dual-layer scanning solution with dual-radio nodes is proposed to scan the monitoring area more intensively. The experiment results show that the efficiency of HAS. HAS achieves very low false positive and false negative rates, and the response time of dual-layer scanning is 54.2% less than single layer scanning.

In wireless sensor networks, the adversary may inject false reports to exhaust network energy or trigger false alarms with compromised sensor nodes. In response to the problems of existing schemes on the security resiliency, applicability and filtering effectiveness, this paper proposes a scheme, referred to as Grouping-enhanced Resilient Probabilistic En-route Filtering (GRPEF). In GRPEF, an efficient distributed algorithm is proposed to group nodes without incurring extra groups, and a multiaxis division based approach for deriving location-aware keys is used to overcome the threshold problem and remove the dependence on the sink immobility and routing protocols. Compared to the existing schemes, GRPEF significantly improves the effectiveness of the en-route filtering and can be applied to the sensor networks with mobile sinks while reserving the resiliency.

In-network aggregation provides an energy-efficient way to extract summarization information from sensor networks. Continuous aggregation is usually required in many sensor applications to obtain the temporal variation information of some interesting aggregates. However, for the continuous in-network aggregation in a hostile environment, the adversary could manipulate a series of aggregation results through compromised nodes to fabricate false temporal variation patterns of the aggregates. Existing secure aggregation schemes conduct one individual verification for each aggregation result. Due to the high rate and the long period of a continuous aggregation, directly applying these schemes to detect false temporal variation pattern would incur a great communication cost. In this paper, we identify distinct design issues for protecting continuous in-network aggregation and propose a novel scheme to detect false temporal variation patterns. Compared with the existing schemes, our scheme greatly reduces the communication cost by selecting and checking only a small part of aggregation results to verify the correctness of the temporal variation patterns in a time window. The checking of the aggregation results uses a sampling-based approach, which enables our scheme independent of any particular in-network aggregation protocol. We also propose a series of security mechanisms to protect the sampling process. Both theoretical analysis and simulations show the effectiveness and efficiency of our scheme.

Aggregations of sensed data are very important for users to get summary information about monitored area in applications of wireless sensor networks (WSNs). As the approximate aggregation results are enough for users to perform analysis and make decisions, many approximate aggregation algorithms are proposed for WSNs. However, most of the algorithms have fixed error bounds and cannot meet arbitrary precision requirement, the uniform sampling based algorithm which can reach arbitrary precision is just suitable for the static networks. Considering the dynamic property of WSNs, in this paper, we propose an approximate aggregation algorithm based on Bernoulli sampling to satisfy arbitrary precision requirement. Besides, two adaptive algorithms are also proposed, one is for adapting the sample with varying of precision requirement, the other is for adapting the sample with varying of sensed data. The theoretical analysis and experiment results show that the proposed algorithms have high performance in terms of accuracy and energy consumption.

In sensor networks, the adversaries can inject false data reports from compromising nodes. Previous approaches for filtering false reports, notably statistical en-route filtering, adopt a simple strategy for grouping sensor nodes that requires redundant groups and thus decrease the filtering effectiveness. Worse still, they either suffer a threshold problem, which may lead to complete breakdown of the security protection when the threshold is exceeded, or are dependent on sink stationarity and specific routing protocols, which cannot work with mobile sinks and various routing protocols. In response to these, this paper proposes a scheme, referred to as grouping-based resilient statistical en-route filtering (GRSEF), in which nodes are grouped once deployed without requiring redundant groups and a location-aware approach based on terrain division along multiple axes is proposed for key derivation. The design of GRSEF, which is independent of sink stationarity and routing protocols, provides a well suitable en-routing filtering solution for sensor networks with mobile sinks. Analytical and simulation results verify that the scheme significantly improves the filtering effectiveness and efficiently achieves the resiliency against node compromise.

A wireless sensor network is often deployed for environment monitoring and event inspection. Among these applications, the sink of the network usually requires the data generated on each sensor node periodically, and such a network is called a data-gathering sensor network. In each round of the data gathering process, a sensor node sends its reading via a single-hop or multi-hop path to the sink. Because the sensor nodes are usually battery-powered with limited energy, efficient routing strategy is required to reduce and balance the energy consumption of the sensor nodes in data transmission. This paper studies the problem of maximizing the lifetime of a data-gathering sensor network, which is defined as the number of rounds until the first node depletes its energy. We prove that the problem is NP-complete, and then formulate it as an integer program to get close to optimality. We further propose a polynomial-time and provably near optimal algorithm to reduce the tremendous computation and storage cost of the integer program. Finally, we evaluate the efficiency of our algorithms by extensive experiments.

Network monitoring is a basic component for intrusion detection and also an energy-expensive task. However in the existing works the contradiction between energy efficiency and security of network monitoring isnpsilat well handled. In this paper, we propose SpyMon, a network monitoring mechanism for the sensor network. To achieve energy-efficiency and reliability, a subset of sensor nodes are randomly selected as monitors in the network and each sensor node is monitored by at least k nodes. For the resistance against the attacks, the monitors are protected from identity exposure to prevent them becoming the explicit targets of adversaries. A collective monitoring triggering scheme is also proposed to further improve the capability and reliability of monitoring. Our analysis shows that SpyMon is resilient against node compromise while attaining energy efficiency.