Relay Stations (RSs) can be deployed in a wireless network to extend its coverage and improve its capacity. Smart (directional) antennas can enhance the functionalities of RSs by forming one or multiple beams only towards intended receivers. In this paper, we focus on the topology control approach for efficient communications in wireless relay networks with smart antennas. This approach precomputes an antenna pattern for each node such that an efficient network topology can be formed for future communications. The corresponding optimization problem is formally defined as the Beam Selection Problem (BSP). First, we present an Integer Linear Programming (ILP) formulation to provide optimal solutions. Then we present a Linear Programming (LP) rounding-based algorithm for the BSP and show it has a constant factor approximation ratio. We also present a simple and fast greedy algorithm to solve the problem. Extensive simulation results show that the proposed algorithms provide close-to-optimal performance.
As more and more data-intensive applications have been moved to the cloud, the cloud network has become the new performance bottleneck for cloud applications. To boost application performance, the concept of coflow has been proposed to bring application-awareness into the cloud network. A coflow consists of many individual data flows, and a coflow is completed only when all its component flows are transmitted. The network performance of a cloud application is dependent on the completion time of coflows, rather than the completion time of each individual flow. Existing coflow-aware optimization solutions employ flow preemption to reduce the completion time, which brings difficulty in practical implementation and non-negligible overhead. In this paper, we study the non-preemptive coflow scheduling and routing problem in the cloud network. We propose an offline optimization framework for coflow scheduling, as well as two subroutines for coflow routing using single-path routing and multi-path routing respectively. We also show that our proposed framework is easily extensible to the online scenario. Extensive evaluations show that the proposed solutions can greatly reduce coflow completion time compared to coflow-agnostic solutions, and are also computationally efficient.
Visual Question Answering (VQA) has emerged as one of the most challenging tasks in artificial intelligence due to its multi-modal nature. However, most existing VQA methods are incapable of handling Knowledge-based Visual Question Answering (KB-VQA), which requires external knowledge beyond visible contents to answer questions about a given image. To address this issue, we propose a novel framework that endows the model with capabilities of answering more general questions, and achieves a better exploitation of external knowledge through generating Multiple Clues for Reasoning with Memory Neural Networks (MCR-MemNN). Specifically, a well-defined detector is adopted to predict image-question related relation phrases, each of which delivers two complementary clues to retrieve the supporting facts from external knowledge base (KB), which are further encoded into a continuous embedding space using a content-addressable memory. Afterwards, mutual interactions between visual-semantic representation and the supporting facts stored in memory are captured to distill the most relevant information in three modalities (i.e., image, question, and KB). Finally, the optimal answer is predicted by choosing the supporting fact with the highest score. We conduct extensive experiments on two widely-used benchmarks. The experimental results well justify the effectiveness of MCR-MemNN, as well as its superiority over other KB-VQA methods.
Video anomaly detection is commonly used in many applications such as security surveillance and is very challenging.A majority of recent video anomaly detection approaches utilize deep reconstruction models, but their performance is often suboptimal because of insufficient reconstruction error differences between normal and abnormal video frames in practice. Meanwhile, frame prediction-based anomaly detection methods have shown promising performance. In this paper, we propose a novel and robust unsupervised video anomaly detection method by frame prediction with proper design which is more in line with the characteristics of surveillance videos. The proposed method is equipped with a multi-path ConvGRU-based frame prediction network that can better handle semantically informative objects and areas of different scales and capture spatial-temporal dependencies in normal videos. A noise tolerance loss is introduced during training to mitigate the interference caused by background noise. Extensive experiments have been conducted on the CUHK Avenue, ShanghaiTech Campus, and UCSD Pedestrian datasets, and the results show that our proposed method outperforms existing state-of-the-art approaches. Remarkably, our proposed method obtains the frame-level AUROC score of 88.3% on the CUHK Avenue dataset.
Library-based record and replay tools aim to reproduce an application's execution by recording the results of selected functions in a log and during replay returning the results from the log rather than executing the functions. These tools must ensure that a replay run is identical to the record run. The challenge in doing so is that only invocations of a function by the application should be recorded, recording the side effects of a function call can be difficult, and not executing function calls during replay, multithreading, and the presence of the tool may change the application's behavior from recording to replay. These problems have limited the use of such tools.R2 allows developers to choose functions that can be recorded and replayed correctly. Developers annotate the chosen functions with simple keywords so that R2 can handle calls with side effects andmultithreading. R2 generates code for record and replay from templates, allowing developers to avoid implementing stubs for hundreds of functions manually. To track whether an invocation is on behalf of the application or the implementation of a selected function, R2 maintains a mode bit, which stubs save and restore.We have implemented R2 on Windows and annotated large parts (1,300 functions) of the Win32 API, and two higher-level interfaces (MPI and SQLite). R2 can replay multithreaded web and database servers that previous library-based tools cannot replay. By allowing developers to choose high-level interfaces, R2 can also keep recording overhead small; experiments show that its recording overhead for Apache is approximately 10%, that recording and replaying at the SQLite interface can reduce the log size up to 99% (compared to doing so at the Win32 API), and that using optimization annotations for BitTorrent and MPI applications achieves log size reduction ranging from 13.7% to 99.4%.
Electric vehicles (EVs) are considered to be a promising solution for current gas shortage and emission problems. To maximize the benefits of using EVs, regulated and optimized charging control needs to be provided by load aggregators for connected vehicles. An EV charging network is a typical cyber-physical system, which includes a power grid and a large number of EVs and aggregators that collect information and control the charging procedure. In this paper, we studied EV charging scheduling problems from a customer's perspective by jointly considering the aggregator's revenue and customers' demands and costs. We considered two charging scenarios: static and dynamic. In the static charging scenario, customers' charging demands are provided to the aggregator in advance; however, in the dynamic charging scenario, an EV may come and leave at any time, which is not known to the aggregator in advance. We present linear programming (LP)-based optimal schemes for the static problems and effective heuristic algorithms for the dynamic problems. The dynamic scenario is more realistic; however, the solutions to the static problems can be used to show potential revenue gains and cost savings that can be brought by regulated charging and, thus, can serve as a benchmark for performance evaluation. It has been shown by extensive simulation results based on real electricity price and load data that significant revenue gains and cost savings can be achieved by optimal charging scheduling compared with an unregulated baseline approach, and moreover, the proposed dynamic charging scheduling schemes provide close-to-optimal solutions.
Purpose Drawing upon the theory of planned behavior (TPB) and the self-regulation framework, the purpose of this paper is to investigate whether and how factors for social media continuance behaviors work differently between social networking sites and microblogging. Design/methodology/approach A survey method was used to collect two samples of 557 social networking sites users and 568 microblogging users. The proposed research model was tested with the structural equation modeling technique. Findings The empirical results demonstrate that the impacts of influencing factors on users’ continuance behaviors vary by types of social media services. Information sharing has a stronger impact on microblog users’ satisfaction than social network users while social interaction has a stronger impact on satisfaction for social network users than microblog users. In addition, interpersonal influence is more effective in shaping satisfaction for the social network users while media influence is more effective in shaping satisfaction for the microblog users. Originality/value This is one of the first studies that integrate TPB with Bagozzi’s self-regulation framework to understand the behavioral model of social networking and microblogging continuance. The findings show that the impacts of attitudinal beliefs regarding information sharing and social interaction on social media users’ satisfaction are different across social networking and microblogging contexts. Moreover, this study also reveals different effects of two specific subjective norms – interpersonal and media influence – on continued use of social networking and microblogging.
Approach and avoidance are two major types of behavioral responses when consumers encounter interferences caused by online advertising. This paper argues that approach-avoidance is not the only dimension from which researchers can examine behavioral responses toward online advertising. The inclusion of the active-passive behaviors dimension enriches the understanding of consumers’ coping strategies. Active and passive behaviors differ from each other by the intensity of coping efforts. Active behavioral responses imply that consumers act upon online ads and make efforts to approach or avoid them. Passive behavioral responses indicate that consumers make little efforts to change the current status, and would rather approach or avoid in a passive way. Data was collected through an online survey by asking participants to recall their experiences with online ads and their behavioral responses. We found that the effects of ad design characteristics (content, form, and behavior) on consumers’ behavioral responses differ across two-dimensions: Approach-Avoidance and Active-Passive. In addition, these effects also vary when consumers have different views (negative vs. positive) of the online ads. The contribution of this study lies in suggesting the two-dimensional view of studying consumers’ responses toward online ads and in deepening our understanding of consumer behavior in dealing with digital artefacts in general.
In this paper, we aim to study networking problems from a whole new perspective by leveraging emerging deep learning, to develop an experience-driven approach, which enables a network or a protocol to learn the best way to control itself from its own experience (e.g., runtime statistics data), just as a human learns a skill. We present design, implementation and evaluation of a deep reinforcement learning (DRL)-based control framework, DRL-CC (DRL for Congestion Control), which realizes our experience-driven design philosophy on multi-path TCP (MPTCP) congestion control. DRL-CC utilizes a single (instead of multiple independent) agent to dynamically and jointly perform congestion control for all active MPTCP flows on an end host with the objective of maximizing the overall utility. The novelty of our design is to utilize a flexible recurrent neural network, LSTM, under a DRL framework for learning a representation for all active flows and dealing with their dynamics. Moreover, we, for the first time, integrate the above LSTM-based representation network into an actor-critic framework for continuous (congestion) control, which leverages the emerging deterministic policy gradient to train critic, actor, and LSTM networks in an end-to-end manner. We implemented DRL-CC based on the MPTCP implementation in the Linux kernel. The experimental results show that 1) DRL-CC consistently and significantly outperforms a few well-known MPTCP congestion control algorithms in terms of goodput without sacrificing fairness, 2) it is flexible and robust to highly-dynamic network environments with time-varying flows, and 3) it is friendly to regular TCP.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
