Federated learning (FL) is a promising paradigm that can enable collaborative model training between vehicles while protecting data privacy, thereby significantly improving the performance of intelligent transportation systems (ITSs). In vehicular networks, due to mobility, resource constraints, and the concurrent execution of multiple training tasks, how to allocate limited resources effectively to achieve optimal model training of multiple tasks is an extremely challenging issue. In this paper, we propose a mobility-aware multi-task decentralized federated learning (MMFL) framework for vehicular networks. By this framework, we address task scheduling, subcarrier allocation, and leader selection, as a joint optimization problem, termed TSLP. For the case with a single FL task, we derive the convergence bound of model training. For general cases, we first model TSLP as a resource allocation game, and prove the existence of a Nash equilibrium (NE). Then, based on this proof, we reformulate the game as a decentralized partially observable Markov decision process (DEC-POMDP), and develop an algorithm based on heterogeneous-agent proximal policy optimization (HAPPO) to solve DEC-POMDP. Finally, numerical results are used to demonstrate the effectiveness of the proposed algorithm.

WiFi-based human activity recognition (HAR) holds significant application potential across various fields. To handle dynamic environments where new activities are continuously introduced, WiFi-based HAR systems must adapt by learning new concepts without forgetting previously learned ones. Furthermore, retaining knowledge from old activities by storing historical exemplar is impractical for WiFi-based HAR due to privacy concerns and limited storage capacity of edge devices. In this work, we propose ConSense, a lightweight and fast-adapted exemplar-free class incremental learning framework for WiFi-based HAR. The framework leverages the transformer architecture and involves dynamic model expansion and selective retraining to preserve previously learned knowledge while integrating new information. Specifically, during incremental sessions, small-scale trainable parameters that are trained specifically on the data of each task are added in the multi-head self-attention layer. In addition, a selective retraining strategy that dynamically adjusts the weights in multilayer perceptron based on the performance stability of neurons across tasks is used. Rather than training the entire model, the proposed strategies of dynamic model expansion and selective retraining reduce the overall computational load while balancing stability on previous tasks and plasticity on new tasks. Evaluation results on three public WiFi datasets demonstrate that ConSense not only outperforms several competitive approaches but also requires fewer parameters, highlighting its practical utility in class-incremental scenarios for HAR.

Vertical Federated Learning (VFL) is a framework that partitions data based on features. In real-world scenarios, many participants often have limited computational resources, making global synchronization inefficient and time-consuming. Asynchronous VFL addresses this by allowing participants to update their models with potentially outdated information, which accelerates the training process but can introduce instability. In this paper, we propose a novel asynchronous VFL algorithm that leverages buffer storage for embeddings, effectively balancing the strengths of both synchronous and asynchronous approaches. We provide a theoretical analysis of the algorithm’s convergence properties, ensuring that it maintains robust performance despite the use of stale updates. Importantly, our method preserves data locality by eliminating the need to share original features or local model parameters among participants. Finally, experimental results show that our method significantly improves both convergence speed and model performance compared to existing approaches.

Deep Learning methodologies have demonstrated exceptional efficacy in sentiment classification tasks. However, their extended inference times often impede practical deployment, particularly in resource-constrained environments. This paper addresses the challenge of reducing inference time by introducing a novel in-GPU caching approach, termed SentireCache, specifically designed for sentiment classification tasks. While traditional caching methods with the cosine similarity measurement have shown some reduction in inference time, they suffer from low hit rates and accuracy. To overcome this limitation, we incorporate a token filtering mechanism based on saliency into the caching system, along with simplified similarity calculation methods. The effectiveness of our proposed approach is theoretically analyzed. Moreover, extensive experimentation is conducted to compare SentireCache with other state-of-the-art caching methods. The results demonstrate a significant 37.7% reduction in inference time with an average performance degradation of 4.69%.

Federated learning (FL) has garnered increased attention in the field of distributed machine learning and privacy computing. In the FL setup, effective and efficient unlearning algorithms are required to remove the impact of specific training data from the trained model, called federated unlearning. However, traditional machine unlearning algorithms face limitations in FL systems because the client data is private and even non-IID. In this paper, we propose a new federated unlearning algorithm called FednP. Our approach involves dividing the client set into subsets using multiple different partitions. We then train constituent models for each client subset within these partitions using existing FL algorithms and aggregate the results of constituent models for predictions. With multiple partitions, FednP limits the influence of the data to be erased within its belonging subsets, while it also improves the accuracy of the aggregated prediction. Based on the multiple-partition framework, we design partition creation methods to effectively enhance the prediction accuracy. Furthermore, we propose a cost reduction method to reduce the cost of training/retraining. Our extensive experiments on various datasets and model architectures demonstrate that FednP improves prediction accuracy while well-controls the additional cost.

As a promising approach to deal with distributed data, Federated Learning (FL) achieves major advancements in recent years. FL enables collaborative model training by exploiting the raw data dispersed in multiple edge devices. However, the data is generally non-independent and identically distributed, i.e., statistical heterogeneity, and the edge devices significantly differ in terms of both computation and communication capacity, i.e., system heterogeneity. The statistical heterogeneity leads to severe accuracy degradation while the system heterogeneity significantly prolongs the training process. In order to address the heterogeneity issue, we propose an Asynchronous Staleness-aware Model Update FL framework, i.e., FedASMU, with two novel methods. First, we propose an asynchronous FL system model with a dynamical model aggregation method between updated local models and the global model on the server for superior accuracy and high efficiency. Then, we propose an adaptive local model adjustment method by aggregating the fresh global model with local models on devices to further improve the accuracy. Extensive experimentation with 6 models and 5 public datasets demonstrates that FedASMU significantly outperforms baseline approaches in terms of accuracy (0.60% to 23.90% higher) and efficiency (3.54% to 97.98% faster).

In recent years, federated learning has emerged as an effective approach for the collaborative learning of decentralized data. Vertical federated learning (VFL) is a scenario of federated learning for cross-silo cooperation, where multiple parties with different features about the same set of data jointly train machine learning models without exposing their raw data. Most existing works of VFL focus on a single-job training of one machine learning model. In this paper, we propose a new framework for multi-job VFL, where multiple independent models are trained simultaneously in a cross-silo environment. We formulate the multi-job VFL scheduling problem and propose an efficient solution based on the rolling horizon method. We conduct extensive experiment to evaluate the performance of the solution. The experimental results show that our algorithm outperforms the other baseline algorithms.

Federated learning (FL) has recently received more and more attention in the joint field of distributed machine learning (ML) and privacy computing. Similar to the traditional ML systems, there exists the need of effective and efficient unlearning algorithms to unlearn certain training data from the FL model. The traditional machine unlearning algorithms have limitations for the FL systems, since the data of clients are both private and non-IID. In this paper, we propose a new algorithm for federated unlearning called FedUMP to improve the model performance and accelerate the unlearning process. Its main idea is to first create multiple different client partition strategies, each of which divides the clients into several subsets. Then we independently train subset models for all client subsets and aggregate the results of subset models for predictions. Furthermore, we propose a retraining acceleration method to reduce the time consumption with multiple partitions, and a partition strategy design method to search for good partition strategies efficiently. Extensive experiments on various datasets and model architectures demonstrate that FedUMP improves both model performance and unlearning speed.

The rapid development of artificial intelligence has led to a continuous expansion in the scale of the neural network models being trained, resulting in the gradual insufficiency of model training resources to support the increasingly larger models. Electric vehicles have become more intelligent thanks to their increasingly powerful computing resources. When these vehicles are not in operation, a significant amount of computing resources are left idle. This paper proposes a new distributed machine-learning system, i.e., virtual data center with parking electric vehicles (PDC), which is composed of central servers, hybrid local area networks, and electric vehicles in parking lots. The PDC is aimed at utilizing the powerful computational capabilities of electric vehicles in parking lots to provide robust computational resources for model training in favor of organizations such as universities and companies. This is accomplished by collecting jobs that need to be trained from the server, gathering information about the computation and communication resources of parking lots, and distributing the jobs that need to be trained to various electric vehicles. We further propose a reinforcement learning-based scheduling algorithm to minimize the job completion time for the jobs. The simulation results justify the feasibility of the PDC system and the efficiency of our proposed scheduling algorithm.

The integration of Artificial Intelligence (AI) and Internet of Things (IoT) devices has given rise to IoAT, promising transformative applications across various domains. Federated Learning (FL) and Split Learning (SL) are pivotal in harnessing the potential of IoAT, enabling decentralized model training while preserving data privacy. However, the heterogeneity and scalability challenges in IoAT environments necessitate advanced frameworks. In this paper, we introduce an integration of block-wise Neural Architecture Search (NAS) with a multi-partition SFL framework, called NASFLY. This approach uses only devices for actual model training and offers flexible scaling of model fragments to accommodate a wide range of device capabilities. In particular, NASFLY allows devices to utilize idle periods during the lengthy SFL forward and backward propagation phases. This is achieved by employing auxiliary model components dispatched from the server to conduct local supernet elastification using the device’s local dataset. Our method alternates between SFL for backbone network optimization and the local supernet elastification within NAS, where knowledge from the backbone network is transferred to the local supernet branches using distillation techniques. We also propose a device clustering algorithm to further improve training efficiency. Our experimental results demonstrate that this methodology significantly enhances device utilization and improves training efficiency compared with the conventional SFL.

Summary While data is distributed in multiple edge devices, federated learning (FL) is attracting more and more attention to collaboratively train a machine learning model without transferring raw data. FL generally exploits a parameter server and a large number of edge devices during the whole process of the model training, while several devices are selected in each round. However, straggler devices may slow down the training process or even make the system crash during training. Meanwhile, other idle edge devices remain unused. As the bandwidth between the devices and the server is relatively low, the communication of intermediate data becomes a bottleneck. In this article, we propose time‐efficient asynchronous federated learning with sparsification and quantization, that is, TEASQ‐Fed. TEASQ‐Fed can fully exploit edge devices to asynchronously participate in the training process by actively applying for tasks. We utilize control parameters to choose an appropriate number of parallel edge devices, which simultaneously execute the training tasks. In addition, we introduce a caching mechanism and weighted averaging with respect to model staleness to further improve the accuracy. Furthermore, we propose a sparsification and quantitation approach to compress the intermediate data to accelerate the training. The experimental results reveal that TEASQ‐Fed improves the accuracy (up to 16.67% higher) while accelerating the convergence of model training (up to twice faster).

Despite achieving remarkable performance, Federated Learning (FL) encounters two important problems, i.e., low training efficiency and limited computational resources. In this article, we propose a new FL framework, i.e., FedDUMAP, with three original contributions, to leverage the shared insensitive data on the server in addition to the distributed data in edge devices so as to efficiently train a global model. First, we propose a simple dynamic server update algorithm, which takes advantage of the shared insensitive data on the server while dynamically adjusting the update steps on the server in order to speed up the convergence and improve the accuracy. Second, we propose an adaptive optimization method with the dynamic server update algorithm to exploit the global momentum on the server and each local device for superior accuracy. Third, we develop a layer-adaptive model pruning method to carry out specific pruning operations, which is adapted to the diverse features of each layer so as to attain an excellent tradeoff between effectiveness and efficiency. Our proposed FL model, FedDUMAP, combines the three original techniques and has a significantly better performance compared with baseline approaches in terms of efficiency (up to 16.9 times faster), accuracy (up to 20.4% higher), and computational cost (up to 62.6% smaller).

Tiny machine learning (TinyML) has emerged recently for resource constrained Internet of Things (IoT) devices. However, the deployed TinyML model cannot handle outof-distribution (OOD) inputs appropriately. While many high-accuracy OOD detection methods have emerged, they often ignore the limitations of the deployment environment. In this paper, we propose a novel effective out-of-distribution detection method for TinyML (TinyOOD), which exploits cascading early exit and channel-attention-based neural mean discrepancy (CA-NMD) for dynamic and efficient OOD detection on microcontroller units (MCUs). To demonstrate its effectiveness, we extensively evaluate TinyOOD using four public datasets, one as the in-distribution (ID) dataset and the others as the OOD datasets. Experiments demonstrate that TinyOOD significantly reduces the computations by up to 38.23% in inference while maintaining the performance of OOD detection.

Federated learning is a distributed machine learning framework that enables distributed edge devices to jointly train machine learning models without exchanging private data. Because federated learning can leverage edge devices’ data and computing resources, the research on federated learning has set off an upsurge. However, one of major challenges faced by federated learning is the limited computation and communication resources of edge devices, which leads to low efficiency of federated learning. In this paper, we propose a federated learning framework FedPipeline, which focuses on the local training and uploading process to improve the convergence efficiency. Different from the existing federated learning algorithms with disjoint sequential steps of training and uploading for devices, FedPipeline exploits the parallelism of training and uploading that devices upload part of the model updates in advance while they are conducting local training simultaneously. In this way, the computation and communication resources of devices can be better utilized. We further design strategies for the scheduling of early uploading and the selection of uploaded updates to maximize the benefit of FedPipeline. Experimental results show that FedPipeline can effectively accelerate the convergence speed of federated learning and preserve the accuracy of the final model.

With the fast development of modern networking technologies, the transmission rate and reliability of wireless networks have been greatly improved. Meanwhile, the fast-developing Internet of Things (IoT) devices provide continually increasing computation capability. Federated learning (FL) was proposed to leverage communication and computation resources to perform machine learning (ML) tasks on IoT devices. In wireless FL systems, devices train ML models with local data sets, and a base station (BS) aggregates these trained models so that data privacy is guaranteed by the isolation of data sets. However, in the existing works, the same global model is transmitted to devices individually over the wireless channel multiple times, while the updated local models are received only by the BS, which ignores the wireless broadcast channel, incurs large communication overhead, and slows down the overall training process. In this article, we propose the FedBroadcast protocol to efficiently exploit the shared wireless broadcast channel for the two-way model transmission in FL. In the downloading step, we let BS broadcast the global model once for all scheduled devices, and design a dynamic programming-based algorithm to schedule devices optimally. In the uploading step, we also leverage the wireless broadcast channel and select some devices to receive all the updated local models without waiting for model downloading in the next round. Finally, to solve the potential block-cyclic sampling problem brought by device scheduling, we adopt pluralistic averaging modification, which improves the convergence performance under extreme data distributions. Extensive experiments demonstrate that FedBroadcast outperforms the existing wireless FL under different system settings.

Recent years have witnessed a large amount of decentralized data in various (edge) devices of end-users, while the decentralized data aggregation remains complicated for machine learning jobs because of regulations and laws. As a practical approach to handling decentralized data, Federated Learning (FL) enables collaborative global machine learning model training without sharing sensitive raw data. The servers schedule devices to jobs within the training process of FL. In contrast, device scheduling with multiple jobs in FL remains a critical and open problem. In this article, we propose a novel multi-job FL framework, which enables the training process of multiple jobs in parallel. The multi-job FL framework is composed of a system model and a scheduling method. The system model enables a parallel training process of multiple jobs, with a cost model based on the data fairness and the training time of diverse devices during the parallel training process. We propose a novel intelligent scheduling approach based on multiple scheduling methods, including an original reinforcement learning-based scheduling method and an original Bayesian optimization-based scheduling method, which corresponds to a small cost while scheduling devices to multiple jobs. We conduct extensive experimentation with diverse jobs and datasets. The experimental results reveal that our proposed approaches significantly outperform baseline approaches in terms of training time (up to 12.73 times faster) and accuracy (up to 46.4% higher).

Recent years have witnessed a large amount of decentralized data in multiple (edge) devices of end-users, while the aggregation of the decentralized data remains difficult for machine learning jobs due to laws or regulations. Federated Learning (FL) emerges as an effective approach to handling decentralized data without sharing the sensitive raw data, while collaboratively training global machine learning models. The servers in FL need to select (and schedule) devices during the training process. However, the scheduling of devices for multiple jobs with FL remains a critical and open problem. In this paper, we propose a novel multi-job FL framework to enable the parallel training process of multiple jobs. The framework consists of a system model and two scheduling methods. In the system model, we propose a parallel training process of multiple jobs, and construct a cost model based on the training time and the data fairness of various devices during the training process of diverse jobs. We propose a reinforcement learning-based method and a Bayesian optimization-based method to schedule devices for multiple jobs while minimizing the cost. We conduct extensive experimentation with multiple jobs and datasets. The experimental results show that our proposed approaches significantly outperform baseline approaches in terms of training time (up to 8.67 times faster) and accuracy (up to 44.6% higher).

Despite achieving remarkable performance, Federated Learning (FL) suffers from two critical challenges, i.e., limited computational resources and low training efficiency. In this paper, we propose a novel FL framework, i.e., FedDUAP, with two original contributions, to exploit the insensitive data on the server and the decentralized data in edge devices to further improve the training efficiency. First, a dynamic server update algorithm is designed to exploit the insensitive data on the server, in order to dynamically determine the optimal steps of the server update for improving the convergence and accuracy of the global model. Second, a layer-adaptive model pruning method is developed to perform unique pruning operations adapted to the different dimensions and importance of multiple layers, to achieve a good balance between efficiency and effectiveness. By integrating the two original techniques together, our proposed FL model, FedDUAP, significantly outperforms baseline approaches in terms of accuracy (up to 4.8% higher), efficiency (up to 2.8 times faster), and computational cost (up to 61.9% smaller).

Federated learning (FL) has attracted more and more attention recently. The integration of FL and edge computing makes the edge system more efficient and intelligent. FL usually uses the server to actively select certain edge devices to participate in the global model training. However, the selected edge devices may be stragglers, or even crash during training. Meanwhile, the unselected idle edge devices cannot be fully utilized for training. Therefore, besides the widely studied communication efficiency and data heterogeneity issues in FL, we also take the above time efficiency into consideration, and propose a time-efficient asynchronous federated learning protocol, TEA-Fed, to solve these problems. With TEA-Fed, idle edge devices actively apply for training tasks and participate in model training asynchronously once assigned tasks. Considering that there may be a huge number of edge devices in edge computing, we introduce control parameters to limit the number of devices participating in training the identical model at the same time. Meanwhile, we also introduce caching mechanism and weighted averaging with respect to model staleness in the model aggregation step to reduce the adverse effects of model staleness and further improve the accuracy of the global model. Finally, the experimental results show that the protocol can accelerate the convergence of model training, improve the accuracy, and has robustness to heterogeneous data.

Summary Currently, Mobile edge computing (MEC) is facing a great challenge that is how to make full use of edge resources to provide a seamless support for compute‐intensive latency‐sensitive applications. Prior studies often make a simple assumption that tasks can be executed upon every edge server, but the assumption does not hold in practical scenarios. Because a specific application task often corresponds to a certain service that provides the corresponding running environment, whereas an edge server only has limited resources and cannot offer too many services. How to decide service deployment of so many types of services among multiple edge servers is also a big challenge. To address the challenge, we study dynamic service deployment for latency‐sensitive applications. We first model the long‐term budget‐constrained latency minimization problem as a multi‐slot latency minimization problem based on the Lyapunov framework. By doing this, the hardness of a problem is significantly reduced, since we never require future information to solve the long‐term optimization. Furthermore, we extend our study by joining the task scheduling optimization, where every edge server is fully utilized in an even more efficient collaborative manner. Our extensive experiments show that the proposed algorithms can bring short latency with low cost.

In cognitive radio networks (CRNs), secondary users need to first discover neighbours and form communication links, referred to as the rendezvous process. Rendezvous between any two secondary users can only be achieved on the same channel. However, the nature of the CRN makes this a challenging problem. Specifically in CRN, not only the network is multi- channel, but the channels available at different nodes may be different. While most of the existing works study pair-wise rendezvous and design channel hopping sequence, in this paper, we focus on the performance improvement of the rendezvous process based on the existing channel hopping sequences with multiple users in CRN. We propose a new strategy, called Group Based Strategy (GBS) to achieve the acceleration, which is flexible to incorporate the existing sequence generation algorithms. Our basic idea is to group the encountered users and schedule rendezvous for them. With the purpose to increase rendezvous diversity, other users or groups can join the group if they get the group rendezvous information. Experiments are conducted to evaluate the proposed scheme. Overall, the performance can be improved by more than 50% under symmetric model or asymmetric model using our accelerating strategy.

The continuous demand for content-rich multimedia is pushing for high-speed and secure transmission approaches. In recent years, linear network coding (LNC) has been shown to be a promising technology to improve network throughput, transmission reliability, and information security. In this paper, we study the optimal transmission topology construction and LNC design for a secure multiple-source multicast to deliver the same content with integral link rates, which can be equivalent to the secure multicast problem with a virtual source, i.e., the integer secure virtual-source multicast (ISVM) problem. The objectives of the ISVM problem include the following: 1) satisfy the weakly secure requirements, 2) maximize the secure multicast rate (SMR), and 3) minimize the transmission cost when the SMR is maximized. First, we analyze the necessary and sufficient condition that there exist a transmission topology with integral link rates and a secure LNC that can achieve a given SMR$R$. Then, we model the ISVM problem as an integer linear programming based on the theoretical analysis and design an efficient transmission topology construction algorithm to solve the ISVM problem by utilizing the Lagrangian relaxation and subgradient method. We also analyze the size of finite field required to construct thedeterministic LNCfor a secure virtual-source multicast and the probability that the virtual-source multicast is weakly secure when usingrandom LNCin the ISVM problem. Finally, we design upper and lower bounds for the ISVM problem and conduct extensive simulations to compare the performance of the proposed algorithms with these two bounds.

Summary With the increasing number of users and a huge scale of data, the service providers of Online Social Networks (OSNs) are facing the problem of how to place users' data to multiple servers. Key‐value stores solve the problem based on consistent hashing, and have become a defacto standard. However, random placement manner of hashing cannot preserve social locality, which leads to high intra‐data center traffic and unpredictable response time. Many existing works solve the problem by using graph partitioning algorithms. These works have two drawbacks: First, the social graph is constructed with ordinary pairwise graph that cannot fully reflect multi‐participant interactions often occurring in OSNs. Second, the underlying network topologies of data center have never been considered. This paper investigates the problem of traffic minimization for OSNs data storage. Motivated by maximally preserving both social locality and distance locality, we formulate the problem as two sub‐problems — hypergraph partitioning and partition‐to‐server mapping, and propose a two‐phase data placement (TDP) scheme. Specifically we present two algorithms to solve partition‐to‐server mapping over two widely used network topologies (i.e.,tree and BCube). Evaluations with a large scale Facebook trace show that TDP significantly reduces intra‐data center traffic as well as load balancing across servers. Copyright © 2016 John Wiley & Sons, Ltd.

Summary Personal cloud storage provides users with convenient data access services. Service providers build distributed storage systems by utilizing cloud resources with distributed hash table (DHT), so as to enhance system scalability. Efficient resource provisioning could not only guarantee service performance, but help providers to save cost. However, the interactions among servers in a DHT‐based cloud storage system depend on the routing process, which makes its execution logic more complicated than traditional multi‐tier applications. In addition, production data centers often comprise heterogeneous machines with different capacities. Few studies have fully considered the heterogeneity of cloud resources, which brings new challenges to resource provisioning. To address these challenges, this paper presents a novel resource provisioning model for service providers. The model utilizes queuing network for analysis of both service performance and cost estimation. Then, the problem is defined as a cost optimization with performance constraints. We propose a cost‐efficient algorithm to decompose the original problem into a sub‐optimization one. Furthermore, we implement a prototype system on top of an infrastructure platform built with OpenStack. It has been deployed in our campus network. Based on real‐world traces collected from our system and Dropbox, we validate the efficiency of our proposed algorithms by extensive experiments. Copyright © 2016 John Wiley & Sons, Ltd.

Obtaining spectrum information efficiently is key to cognitive radio networks (CRNs). The database-driven approach has emerged recently as an alternative or supplement for spectrum sensing, and has been quickly adopted by the government and the industry. Within database-driven CRNs, master devices obtain spectrum information by direct connection to a spectrum database, while slave devices can only access spectrum information indirectly via masters. Out-of-band connections with a second network interface which do not depend on the primary spectrum channels can be used for the communication of spectrum information among masters and slaves. Alternatively, the in-band approach completely based on primary spectrum channels can be used, which eliminates the need for out-of-band connections and eases the adoption of the database-driven spectrum sharing. In this paper, we study the in-band bootstrapping process for database-driven multi-hop CRNs, where master/slave devices form a multi-hop networks and slaves need multi-hop communications to obtain spectrum information from the master during bootstrapping. We propose several protocols to reduce the bootstrapping time and protocol overhead. According to the analysis and simulation results, our proposed protocols can greatly improve the performance.

In cognitive radio networks, secondary users need to first discover neighbours and form communication links, referred to as the rendezvous process. Rendezvous between any two secondary users can only be achieved on the same channel. However, spectrum heterogeneity in cognitive radio networks complicates the rendezvous process. While most of the existing works study pair-wise rendezvous and design channel hopping sequence, in this paper we focus on the general rendezvous problem for multiple users where each user needs to discover all of its neighbours. We propose to maintain and exchange rendezvous information among encountered users, and leverage rendezvous information spread within the network to accelerate the rendezvous process. With such an idea, we propose a general message passing based framework for rendezvous protocol design, which is flexible to incorporate the existing sequence generation algorithms. For the framework, we formulate rendezvous problems from the perspective of individual user, prove the NP-completeness, and propose an efficient greedy channel switching algorithm. Based on the framework, we design several rendezvous protocols for single-hop and multi-hop networks. When channel hopping sequence generation algorithms with guaranteed rendezvous between any two users are used, the rendezvous with the proposed protocols can still be guaranteed. Simulations demonstrate that the rendezvous performance is greatly improved.

Femtocell refers to a new class of low-power, low-cost base stations (BSs) which can provide better coverage and improved voice/data Quality of Service (QoS). Hybrid access in two-tier macro-femto network is regarded as the most ideal access control mechanism to enhance overall network performance. But the implementation of hybrid access is hindered by a lack of market that can motivate ACcess Permission (ACP) trading between Wireless Service Providers (WSPs) and private femtocell owners. In this paper, we propose a reverse auction framework for fair and efficient ACP transaction. Unlike strict outcome (the demand of bidder must be fully satisfied) in most of the existing works on auction design, the proposed auction model allows range outcome, in which WSP accepts partial demand fulfillment and femtocell owners makes best-effort selling. We first propose a Vickery-Clarke-Grove (VCG) based mechanism to maximize social welfare. As the VCG mechanism is too time-consuming, we further propose an alternative truthful mechanism (referred to as suboptimal mechanism) with acceptable polynomial computational complexity. The simulation results have shown that the suboptimal mechanism generates almost the same social welfare and the cost for WSP as VCG mechanism.

Nowadays, with the dramatically increased penetration of wireless access, the conflict between spectrum scarcity and under-utilisation is becoming more and more aggravating. A promising technology to tackle such challenge is cognitive radio, of which spectrum sensing is one of the most important functionalities. In this study, the authors consider an essential problem of energy-efficient spectrum sensing in cognitive radio networks. Although most existing works of spectrum sensing mainly focus on determining an optimal sensing time to maximise the detection probability and/or to minimise the false alarm probability, our problem of how to schedule the power-constrained sensor is much more challenging, because of the trade-off among interests of the primary user, secondary user and sensor. The authors formulate it as a non-linear optimisation problem to maximise the sensor lifetime, with necessary constraints of quality and delay of spectrum sensing, and throughput for performance guarantee of primary and secondary users. Moreover, the authors incorporate the distribution information of channel occupancy/vacancy durations into the problem to yield a desirable solution. They propose a novel framework to obtain the optimal energy-efficient periodic scheduling by adopting both non-linear programming and linear programming. Extensive simulation results are provided to validate our theoretical analysis.

Dynamic spectrum sharing in cognitive radio networks can enhance flexibility, and as a result the efficiency of spectrum usage. In this paper, we address the problem of spectrum sharing in a cognitive radio network where multiple primary and secondary strategic-users are involved. In this scenario, primary users (PUs) would like to offer part of their spectrum to secondary users (SUs) to make extra revenue. PUs face a trade-off since the more spectrum that is being shared with SUs, the more Quality of Service (QoS) degradation their own service will suffer. SUs access the Internet through an Access Point (AP). They have to decide their best strategies by taking into account service satisfaction and payment lost. The profit of PUs and SUs is directly related to the bandwidth allocation and price charging through the AP. Considering this competitive relationship, we model the scenario as a noncooperative game, and analyze it by exploring the properties of Nash Equilibrium (NE) point. The simulation results support our theoretic analysis.

Cooperative communication is a promising technique for future wireless networks, which significantly improves link capacity and reliability by leveraging broadcast nature of wireless medium and exploiting cooperative diversity. However, most of existing works investigate its performance theoretically or by simulation. It has been widely accepted that simulations often fail to faithfully capture many real-world radio signal propagation effects, which can be overcome through developing physical wireless network testbeds. In this work, we build a cooperative testbed based on GNU Radio and Universal Software Radio Peripheral (USRP) platform, which is a promising open-source software-defined radio system. Both single-relay cooperation and multi-relay cooperation can be supported in our testbed. Some key techniques are provided to solve the main challenges during the testbed development: e.g., maximum ratio combine in single-relay transmission and synchronized transmission among multiple relays. Extensive experiments are carried out in the testbed to evaluate performance of various cooperative communication schemes. The results show that cooperative transmission achieves significant performance enhancement in terms of link reliability and end-to-end throughput.

Cognitive radio networking is a promising approach to fulfill the future's need for intelligent, high-performance communication and improve the efficiency of overall spectrum utilization. Testbed evaluation of protocols and algorithms is a must for the development of cognitive radio networks. In this paper, we present an integrated testbed framework for cognitive radio networks. The framework includes necessary components for cognitive radio operations: flexible RF front end, software define signal processing, adaptive MAC layer and network layer as well as a cross layer management interface. Such a design eases the cross layer configuration and performance optimization of cognitive protocol stack, while retaining the advantage of modularity. We design a testbed for ad-hoc cognitive radio network.

Cognitive radio has been proposed in recent years to promote spectrum efficiency by exploiting the existence of spectrum holes. The heterogeneity of both spectrum availability and traffic demand among secondary users has brought significant challenge for efficient spectrum allocation in cognitive radio networks. Observing that spectrum resource can be better matched to traffic demand of secondary users with the help of relay nodes, in this paper we propose to utilize cooperative relays to assist the transmission and improve spectrum efficiency. Different from traditional cooperative communication within a single channel, in our scheme a relay node may be selected for a link to bridge the link's source and destination using its different common channels with those two nodes. With these new logical links composed of both direct link and relay link, new routing protocols are needed so that end-to-end performance can be better improved. Therefore, we define a new link cost, relay- aware link cost, which considers several aspects including channel availability, channel condition, channel utilization and potential relays. Based on this link cost, a relay-assisted routing (RAR) protocol is proposed which includes routing discovery and local adjustment. Simulation results demonstrate the effectiveness of the proposed routing scheme.

Cognitive radio has been proposed in recent years to promote the spectrum utilization by exploiting the existence of spectrum holes. The heterogeneity of both spectrum availability and traffic demand in secondary users has brought significant challenge for efficient spectrum allocation in cognitive radio networks. Observing that spectrum resource can be better matched to traffic demand of secondary users with the help of relay node that has rich spectrum resource, in this paper we exploit a new research direction for cognitive radio networks by utilizing cooperative relay to assist the transmission and improve spectrum efficiency. An infrastructure-based secondary network architecture has been proposed to leverage relay-assisted discontiguous OFDM (D-OFDM) for data transmission. In this architecture, relay node will be selected which can bridge the source and the destination using its common channels between those two nodes. With the introduction of cooperative relay, many unique problems should be considered, especially the issue for relay selection and spectrum allocation. We propose a centralized heuristic solution to address the new resource allocation problem. To demonstrate the feasibility and performance of cooperative relay for cognitive radio, a new MAC protocol has been proposed and implemented in a Universal Software Radio Peripheral (USRP)-based testbed. Experimental results show that the throughput of the whole system is greatly increased by exploiting the benefit of cooperative relay.

Spectrum is a critical yet scarce resource and it has been shown that dynamic spectrum access can significantly improve spectrum utilization. To achieve this, it is important to incentivize the primary license holders to open up their under-utilized spectrum for sharing. In this paper we present a secondary spectrum market where a primary license holder can sell access to its unused or under-used spectrum resources in the form of certain fine-grained spectrum-space-time unit. Secondary wireless service providers can purchase such contracts to deploy new service, enhance their existing service, or deploy ad hoc service to meet flash crowds demand. Within the context of this market, we investigate how to use auction mechanisms to allocate and price spectrum resources so that the primary license holder's revenue is maximized. We begin by classifying a number of alternative auction formats in terms of spectrum demand. We then study a specific auction format where secondary wireless service providers have demands for fixed locations (cells). We propose an optimal auction based on the concept of virtual valuation. Assuming the knowledge of valuation distributions, the optimal auction uses the Vickrey-Clarke-Groves (VCG) mechanism to maximize the expected revenue while enforcing truthfulness. To reduce the computational complexity, we further design a truthful suboptimal auction with polynomial time complexity. It uses a monotone allocation and critical value payment to enforce truthfulness. Simulation results show that this suboptimal auction can generate stable expected revenue.

Significant technology progress has been witnessed in the research area of dynamic spectrum access. However, the success of dynamic spectrum access will not be possible without the evolution of an spectrum and service market which is both stable and efficient. In this paper, we investigate the dynamic access spectrum from the economic point of view. We study three-layer spectrum market with spectrum holder, wireless service provider and end users. In a duopoly situation, two wireless service providers participate in bandwidth competition to purchase spectrum and price competition to attract end users, with the aim of maximizing their own profit. We formulate the wireless service providers' competition as a two-stage game. Under general assumptions about the pricing and demand functions, a unique equilibrium is identified as the outcome of the game, which shows the stability of the market. We further evaluate the market efficiency in a special case of symmetric wireless service providers and affine pricing and demand functions. The result shows the efficiency of the equilibrium is of reasonable level even with non-cooperative wireless service providers.

Dynamic spectrum access can significantly improve the spectrum utilization. For wireless service providers, the emergence of dynamic spectrum access brings new opportunities and challenges. The flexible spectrum acquisition gives a particular provider the chance to easily adapt its system capacity to fit end users' demand. However, the competition among several providers for both spectrum and end users complicates the situation. In this paper, we propose a general three-layer spectrum market model for the future dynamic spectrum access system, in which the interaction among spectrum holder, wireless service providers and end users are considered. We study a duopoly situation, where two wireless service providers participate in bandwidth competition in spectrum purchasing and price competition to attract end users, with the aim of maximizing their own profit. We believe we are the first one to explicitly study the relation of these two competitions in dynamic spectrum market. We formulate the wireless service providers' competition as a non-cooperative two-stage game. We first analyze the static game when full information is available for providers. Under general assumptions about the price and demand functions, a unique pure Nash equilibrium is identified as the outcome of the game, which shows the stability of the market. We further evaluate the market efficiency of the equilibrium in a symmetric case, and show that the gap with the social optimal is bounded within a small constant ratio. When the market information is limited, we provide myopically optimal adjustment algorithms for the providers. With such strategies, short term price updating converges to the Nash equilibrium of the given subgame, while long term bandwidth updating converges to a point close to the Nash equilibrium of the full game.

Radio spectrum resource is of fundamental importance for wireless communication. Recent reports show that most available spectrum has been allocated. While some of the spectrum bands (e.g., unlicensed band, GSM band) have seen increasingly crowded usage, most of the other spectrum resources are underutilized. This drives the emergence of open spectrum and dynamic spectrum access concepts, which allow unlicensed users equipped with cognitive radios to opportunistically access the spectrum not used by primary users. Cognitive radio has many advanced features, such as agilely sensing the existence of primary users and utilizing multiple spectrum bands simultaneously. However, in practice such capabilities are constrained by hardware cost. In this paper, we discuss how to conduct efficient spectrum management in ad hoc cognitive radio networks while taking the hardware constraints (e.g., single radio, partial spectrum sensing and spectrum aggregation limit) into consideration. A hardware-constrained cognitive MAC, HCMAC, is proposed to conduct efficient spectrum sensing and spectrum access decision. We identify the issue of optimal spectrum sensing decision for a single secondary transmission pair, and formulate it as an optimal stopping problem. A decentralized MAC protocol is then proposed for the ad hoc cognitive radio networks. Simulation results are presented to demonstrate the effectiveness of our proposed protocol.

Open spectrum systems allow unlicensed secondary users equipped with cognitive radio to opportunistically access the spectrum underutilized by primary users. Cognitive radio has many advanced features, such as agilely sensing the signal of primary users and utilizing multiple spectrum bands simultaneously, etc. However, the capability of practical cognitive radios is constrained by hardware cost, resulting in partial spectrum sensing and spectrum aggregation limit. In this paper, we take such constraints into consideration and investigate efficient spectrum management in ad hoc cognitive networks with single cognitive radio. A hardware-constrained cognitive MAC, HC-MAC, is proposed to conduct accurate spectrum sensing and spectrum access decision. We identify the optimal spectrum sensing decision for a single secondary transmission pair, and formulate it as an optimal stopping problem. A decentralized MAC protocol is then proposed for the ad hoc cognitive network. Simulation results are given to demonstrate the effectiveness of our proposed protocol.

Limited spectrum resources, inefficient spectrum usage and increasing wireless communication necessitates a paradigm shift from the current fixed spectrum management policy to a more flexible one. With the Federal Communications Commission's (FCC) spectrum policy reform, secondary spectrum sharing has become a viable and promising option. In this paper, we study power control for spectrum sharing among secondary users with interference temperature limit (ITL) constraints at measurement points. Each secondary user will adjust its own transmission power to make sure the overall interference at the measurement points does not exceed the required ITL. Under the assumption that each secondary user is selfish and rational while there is only limited coordination between primary users and secondary users, we study the distributed power control scheme for secondary users. In our proposed system model, secondary users generating the highest interference at a measurement point will back off their transmissions if the ITL is exceeded. A non-cooperative power control game among secondary users is proposed to maximize each user's utility. We identify the Nash equilibrium of the proposed game and analyze its property. Simulations are conducted to demonstrate that the proposed solution can achieve a satisfactory performance in terms of the total transmitting rate of all the secondary users.

We are interested in providing a media-on-demand streaming service to a large population of clients using a peer-to-peer approach. Since the demands of different clients are asynchronous and the contents of clients' buffers are continuously changing, finding partners with expected data and collaborating with them for future content delivery are very important and challenging problems. In this paper, we propose a generic buffer-assisted search (BAS) scheme to improve partner search efficiency by reducing the size of index overlay. We have also developed a novel scheduling algorithm based on deadline-aware network coding (DNC) to fully exploit network resources by dynamically adjusting the coding window size. Extensive simulation results demonstrate that BAS can provide a faster response time with less control cost than the existing search methods, and DNC improves the network capacity utilization and provides high streaming quality under different network conditions.

Group communication is becoming increasingly important with the decreasing cost of broadband access and the growing number of Internet users, which requires efficient multi-source multicast support. This paper targets at addressing two fundamental questions related to application-layer multi- source multicast: how many sources can be simultaneously served and how in reality can such a service be effectively provided. Considering the high bandwidth requirement and potential huge control overhead, in this paper we construct a single distribution tree shared by all the sources. Theoretically, we prove that the optimal single shared tree solution can serve at least m -1 sources, where m is the maximum number of sources that can be supported in the system with any number of trees. Practically, we propose a distributed protocol for tree construction and tree refinement so as to approach the theoretical optimal. The simulation results manifest that our proposed heuristic protocol significantly increases the number of sources that can be served in multi-source sessions and the result approaches to the optimal solution.