Abstract: Presents a listing of the editorial board, board of governors, current staff, committee members, and/or society editors for this issue of the publication. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Abstract: Presents a listing of the editorial board, board of governors, current staff, committee members, and/or society editors for this issue of the publication. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Zhixiong Yang;Waheed U. Bajwa;
Pages: 611 - 627 Abstract: Distributed machine learning algorithms enable learning of models from datasets that are distributed over a network without gathering the data at a centralized location. While efficient distributed algorithms have been developed under the assumption of faultless networks, failures that can render these algorithms nonfunctional occur frequently in the real world. This paper focuses on the problem of Byzantine failures, which are the hardest to safeguard against in distributed algorithms. While Byzantine fault tolerance has a rich history, existing work does not translate into efficient and practical algorithms for high-dimensional learning in fully distributed (also known as decentralized) settings. In this paper, an algorithm termed Byzantine-resilient distributed coordinate descent is developed and analyzed that enables distributed learning in the presence of Byzantine failures. Theoretical analysis (convex settings) and numerical experiments (convex and nonconvex settings) highlight its usefulness for high-dimensional distributed learning in the presence of Byzantine failures. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Mojtaba Shirazi;Azadeh Vosoughi;
Pages: 628 - 645 Abstract: In this paper, we consider the problem of bandwidth-constrained distributed estimation of a Gaussian vector with linear observation model. Each sensor makes a scalar noisy observation of the unknown vector, employs a multi-bit scalar quantizer to quantize its observation, and maps it to a digitally modulated symbol. Sensors transmit their symbols over orthogonal-power-constrained fading channels to a fusion center (FC). The FC is tasked with fusing the received signals from sensors and estimating the unknown vector. We derive the Bayesian Fisher Information Matrix (FIM) for three types of receivers: (i) coherent receiver; (ii) noncoherent receiver with known channel envelopes; and (iii) noncoherent receiver with known channel statistics only. We also derive the Weiss-Weinstein bound (WWB). We formulate two constrained optimization problems, namely maximizing trace and log-determinant of Bayesian FIM under network transmit power constraint, with sensors' transmit powers being the optimization variables (we refer to as FIM-max schemes). We show that for coherent receiver, these problems are concave. However, for noncoherent receivers, they are not necessarily concave. The solution to the trace of Bayesian FIM maximization problem can be implemented in a distributed fashion, in the sense that each sensor calculates its own transmit power using its local parameters. On the other hand, the solution to the log-determinant of Bayesian FIM maximization problem cannot be implemented in a distributed fashion and the FC needs to find the powers (using parameters of all sensors) and inform the active sensors of their transmit powers. We numerically investigate how the FIM-max power allocation across sensors depends on the sensors observation qualities and physical layer parameters as well as the network transmit power constraint. Moreover, we evaluate the system performance in terms of mean square error (MSE) using the solutions of FIM-max schemes, and-compare it with the solution obtained from minimizing the MSE of the LMMSE estimator (MSE-min scheme), and that of uniform power allocation. These comparisons illustrate that, although the WWB is tighter than the inverse of Bayesian FIM, it is still suitable to use FIM-max schemes, since the performance loss in terms of the MSE of the LMMSE estimator is not significant. Furthermore, comparing the performance of different receivers, our numerical results reveal that coherent receiver and noncoherent receiver with known channel statistics have the best and the worst performance, respectively. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Smruti Ranjan Panigrahi;Niclas Björsell;Mats Bengtsson;
Pages: 646 - 656 Abstract: In this paper, a multi-hypothesis distributed detection technique with non-identical local detectors is investigated. Here, for a global event, some of the sensors/detectors can observe the whole set of hypotheses, whereas the remaining sensors can either see only some aspects of the global event or infer more than one hypothesis as a single hypothesis. Another possible option is that different sensors provide complementary information. The local decisions are sent over a multiple access radio channel so that the data fusion is formed in the air before reaching the decision fusion center (DFC). An optimal energy fusion rule is formulated by considering the radio channel effects and the reliability of the sensors together, and a closed-form solution is derived. A receive beamforming algorithm, based on a modification of Lozano's algorithm, is proposed to equalize the channel gains from different sensors. Sensors with limited detection capabilities are found to boost the overall system performance when they are used along with fully capable sensors. The additional transmit power used by these sensors is compensated by the designed fusion rule and the antenna array gain. Additionally, the DFC, equipped with a large antenna array, can reduce the overall transmit energy consumption without sacrificing the detection performance. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Yuqing Ni;Alex S. Leong;Daniel E. Quevedo;Ling Shi;
Pages: 657 - 668 Abstract: We consider the communication channel pricing and selection problem in a networked control system. To encompass the sequentialized nature of the decision-making process, we use game theory and formulate a Stackelberg game framework, where the server first determines the channel pricing strategy, and the clients then make channel selection decisions. Both single-server-single-client (SSSC) scenario and single-server-multi-client (SSMC) scenario are discussed. The existence of an optimal stationary and deterministic policy for the clients is proved. We show that for the SSSC scenario, the server's optimal pricing strategy in terms of maximizing revenue is to ensure that the client uses the good channel all the time. For the SSMC scenario, it is assumed that the channel price remains invariant. As a consequence, each client has an optimal policy with threshold structure. Some properties of the optimal policy pair for both scenarios are obtained. Simulation results confirm the structure and properties of both the server and clients' optimal strategies. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Sowjanya Modalavalsa;Upendra Kumar Sahoo;Ajit Kumar Sahoo;Sananda Kumar;
Pages: 669 - 683 Abstract: Least squared error cost function based conventional diffusion strategies are not robust against outliers in both the desired and input data. In most of the practical scenarios, both the input and desired data contain impulsive noise along with Gaussian noise. The presence of outliers in the measured (input/desired) data can be treated as the impulsive noise. A novel diffusion generalized rank norm algorithm is developed in this article, which is robust against outliers in both input and desired data. The proposed method does not rely on any prior assumption regarding the distribution of the data. The performance analysis of the proposed algorithm is analyzed using asymptotic linearity relation between null and alternative hypotheses of generalized rank norm gradient. Simulation based experiments are carried out to validate the robustness of the proposed algorithm. Another diffusion high breakdown estimator algorithm is proposed, which attains 50% breakdown in both input and desired data space. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Yanli Yuan;De Wen Soh;Howard H. Yang;Tony Q. S. Quek;
Pages: 684 - 697 Abstract: Learning graph Laplacian matrices plays a crucial role in network analytics when a meaningful graph is not readily available from the datasets. However, graph Laplacian inference is an ill-posed problem, since multiple solutions may exist to associate a graph with the data. Recent papers have exploited signal smoothness or graph sparsity to handle this problem, without considering specific graph topological property such as community structure. Community structure is prevalent in many real-world networks, which can be exploited to learn the data better. In this paper, we propose a framework that learns the graph Laplacians with overlapping community structure, named LOCN (Learning Overlapping Community-based Networks). Our framework encompasses and leverages the community structure information, along with attributes such as sparsity and signal smoothness to capture the intrinsic relationships between data entities, such that the estimated graph can optimally fit the data. Furthermore, the refined graph Laplacian can be leveraged to further improve the detection of network communities. As a result, LOCN can not only learn the graph Laplacian with a good data fit, but also detect the underlying network communities with a high quality. We show that LOCN can achieve good results for both synthetic and real data. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Arun Venkitaraman;Saikat Chatterjee;Peter Händel;
Pages: 698 - 710 Abstract: We propose a kernel regression method to predict a target signal lying over a graph when an input observation is given. The input and the output could be two different physical quantities. In particular, the input may not be a graph signal at all or it could be agnostic to an underlying graph. We use a training dataset to learn the proposed regression model by formulating it as a convex optimization problem, where we use a graph-Laplacian based regularization to enforce that the predicted target is a graph signal. Once the model is learnt, it can be directly used on a large number of test data points one-by-one independently to predict the corresponding targets. Our approach employs kernels between the various input observations, and as a result the kernels are not restricted to be functions of the graph adjacency/Laplacian matrix. We show that the proposed kernel regression exhibits a smoothing effect, while simultaneously achieving noise-reduction and graph-smoothness. We then extend our method to the case when the underlying graph may not be known apriori, by simultaneously learning an underlying graph and the regression coefficients. Using extensive experiments, we show that our method provides a good prediction performance in adverse conditions, particularly when the training data is limited in size and is noisy. In graph signal reconstruction experiments, our method is shown to provide a good performance even for a highly under-determined subsampling. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Benedito J. B. Fonseca;
Pages: 711 - 722 Abstract: When designing a sensor system to detect a target emitter in a region of interest, a fusion rule based on the scan statistic can provide improved detection performance when the emitter is uniformly distributed in the region; however, in many applications, the distribution of the emitter location is unknown. In these cases, a possible approach is to design the system considering a least favorable distribution for the emitter location. It is however difficult to find such a distribution because of the difficulty in computing the probabilities of false alarm (PFA) and detection (PD) of the scan statistic. To circumvent this difficulty, this paper proposes both lower and upper bounds for PD, and presents a procedure to obtain a tight lower bound. With these bounds and procedure, systems that ensure a minimum PD under a PFA constraint can be designed. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Esraa Al-sharoa;Mahmood A. Al-khassaweneh;Selin Aviyente;
Pages: 723 - 738 Abstract: Networks provide a powerful tool to model complex systems where the different entities in the system are presented by nodes and their interactions by edges. With the availability of network-type data, different community detection algorithms have been proposed to investigate the organization of the nodes within these networks. In particular, numerous graph-based community detection algorithms have been developed for static networks. However, most real complex systems vary with time. Consequently, it is important to develop graph-based community detection techniques for temporal networks. In this paper, a new low-rank + sparse estimation based evolutionary spectral clustering approach is proposed to detect and track the community structure in temporal networks. The proposed method decomposes the network into low-rank and sparse parts and obtains smooth cluster assignments by minimizing the subspace distance between consecutive time points. The extracted low-rank adjacency matrix is then used for clustering and the subspaces are defined through spectral embedding. The introduced framework is robust to noise and outliers and can detect the community structure in both binary and weighted temporal networks efficiently without making any prior assumptions about the network structure. The proposed approach is evaluated on several weighted and binary simulated and real temporal networks. The results show that the proposed algorithm can detect and track the correct community structure over time efficiently and outperforms state-of-the-art algorithms. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Zuoen Wang;Jingxian Wu;Jing Yang;Yingli Cao;
Pages: 739 - 752 Abstract: The distributed estimation of a spatially correlated random field with decentralized sensor networks is studied in this paper. Nodes in the network take spatial samples of the random field, then each node estimates the values of arbitrary points on the random field by iteratively exchanging information with each other. The objective is to minimize the estimation mean squared error (MSE) while ensuring all nodes reach a distributed consensus on the estimation results. We propose a distributed iterative linear minimum mean squared error (LMMSE) algorithm that contains a state consensus stage and a local estimation stage in each iteration. The proposed algorithm requires the knowledge of the second-order statistics of the random field, and they are estimated by using a distributed learning algorithm with the help of distributed consensus. The key parameters of the algorithm, including an edge weight matrix and a sample weight matrix, are designed to minimize an MSE upper bound at all nodes when the number of iterations is large. It is shown that the optimum performance can be achieved by distributively mapping the high dimension measurement samples from all nodes into a low dimension subspace related to the covariance matrices of data and noise samples. The low-dimension mapping is achieved in a distributed manner through iterative information propagation. The low dimension mapping can significantly reduce the amount of data exchanged in the network, thus improve the convergence speed of the iterative algorithm. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Bicheng Ying;Kun Yuan;Ali H. Sayed;
Pages: 753 - 767 Abstract: This work derives and analyzes an online learning strategy for tracking the average of time-varying distributed signals by relying on randomized coordinate-descent updates. During each iteration, each agent selects or observes a random entry of the observation vector, and different agents may select different entries of their observations before engaging in a consultation step. Careful coordination of the interactions among agents is necessary to avoid bias and ensure convergence. We provide a convergence analysis for the proposed methods, and illustrate the results by means of simulations. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Paolo Addesso;Vincenzo Matta;
Pages: 768 - 778 Abstract: This work deals with networks of agents that exchange information under communication constraints. As a first contribution, the theory of configuration functions is exploited to obtain a general abstract formulation of the network information as a function of the network constraints. As a second contribution, two classic network paradigms are examined: i) a decentralized architecture with remote fusion center; and ii) a fully-Hat decentralized architecture with local data exchange between neighboring agents. It is shown how these paradigms match well with the general formulation in terms of configuration functions. Finally, the statistical concentration properties of configuration functions are exploited to characterize the information growth rate under both the aforementioned network paradigms, revealing the thermodynamic deterministic behavior that emerges with high probability as the network size scales to infinity. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Gowtham Muniraju;Cihan Tepedelenlioglu;Andreas Spanias;
Pages: 779 - 791 Abstract: A novel distributed algorithm for estimating the maximum of the node initial state values in a network, in the presence of additive communication noise is proposed. Conventionally, the maximum is estimated locally at each node by updating the node state value with the largest received measurements in every iteration. However, due to the additive channel noise, the estimate of the maximum at each node drifts at each iteration and this results in nodes diverging from the true max value. Max-plus algebra is used as a tool to study this ergodic process. The subadditive ergodic theorem is invoked to establish a constant growth rate for the state values due to noise, which is studied by analyzing the max-plus Lyapunov exponent of the product of noise matrices in a max-plus semiring. The growth rate of the state values is upper bounded by a constant which depends on the spectral radius of the network and the noise variance. Upper and lower bounds are derived for both fixed and random graphs. Finally, a two-run algorithm robust to additive noise in the network is proposed and its variance is analyzed using concentration inequalities. Simulation results supporting the theory are also presented. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Authors:
Peng Hao;Xianbin Wang;
Pages: 792 - 806 Abstract: Recent literature has demonstrated the improved data discovery and delivery efficiency gained through applying named data networking (NDN) to a variety of information-centric Internet of Things (IoT) applications. However, from a data security perspective, the development of NDN-IoT raises several new authentication challenges. In particular, NDN-IoT authentication may require per-packet-level signatures, thus imposing intolerably high computational and time costs on the resource-poor IoT end devices. This paper proposes an effective solution by seamlessly integrating the lightweight and unforgeable physical-layer identity (PHY-ID) into the existing NDN signature scheme for the mobile edge computing (MEC)-enabled NDN-IoT networks. The PHY-ID generation exploits the inherent signal-level device-specific radio-frequency imperfections of IoT devices, including the in-phase/quadrature-phase imbalance, and thereby avoids adding any implementation complexity to the constrained IoT devices. We derive the offline maximum entropy-based quantization rule and propose an online two-step authentication scheme to improve the accuracy of the authentication decision making. Consequently, a cooperative MEC device can securely execute the costly signing task on behalf of the authenticated IoT device in an optimal manner. The evaluation results demonstrate 1) elevated authentication time efficiency, 2) robustness to several impersonation attacks including the replay attack and the computation-based spoofing attack, and 3) increased differentiation rate and correct authentication probability by applying our integration design in MEC-enabled NDN-IoT networks. PubDate:
Dec. 2019
Issue No:Vol. 5, No. 4 (2019)
Subscription journal
