Large Deviations Performance of Consensus plus Innovations Distributed Detection With Non-Gaussian Observations

被引:43
作者
Bajovic, Dragana [1 ,2 ]
Jakovetic, Dusan [1 ,2 ]
Moura, Jose M. F. [2 ]
Xavier, Joao [1 ]
Sinopoli, Bruno [2 ]
机构
[1] Univ Tecn Lisboa, IST, ISR, Lisbon, Portugal
[2] Carnegie Mellon Univ, Dept Elect & Comp Engn, Pittsburgh, PA 15213 USA
基金
美国国家科学基金会; 美国安德鲁·梅隆基金会;
关键词
Chernoff information; consensus plus innovations; distributed detection; information flow; large deviations; non-Gaussian distributions; performance analysis; random network; SENSOR NETWORKS; DECENTRALIZED DETECTION; MULTIPLE SENSORS; TOPOLOGY; SQUARES;
D O I
10.1109/TSP.2012.2210885
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
We establish the large deviations asymptotic performance (error exponent) of consensus+innovations distributed detection over random networks with generic (non-Gaussian) sensor observations. At each time instant, sensors 1) combine theirs with the decision variables of their neighbors (consensus) and 2) assimilate their new observations (innovations). This paper shows for general non-Gaussian distributions that consensus+innovations distributed detection exhibits a phase transition behavior with respect to the network degree of connectivity. Above a threshold, distributed is as good as centralized, with the same optimal asymptotic detection performance, but, below the threshold, distributed detection is suboptimal with respect to centralized detection. We determine this threshold and quantify the performance loss below threshold. Finally, we show the dependence of the threshold and of the performance on the distribution of the observations: the asymptotic performance of distributed detectors over the same random network with different observations' distributions, for example, Gaussian, Laplace, or quantized, may be different, even though the asymptotic performance of the corresponding centralized detectors is the same.
引用
收藏
页码:5987 / 6002
页数:16
相关论文
共 35 条
[1]  
Aldosari S., 2006, IEEE International Conference on Acoustics, Speech and Signal Processing, P1061
[2]   Detection in sensor networks: The saddlepoint approximation [J].
Aldosari, Saeed A. ;
Moura, Jose M. F. .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2007, 55 (01) :327-340
[3]  
[Anonymous], 1993, Large deviations techniques and applications
[4]  
[Anonymous], 1993, Probability
[5]  
Bajovic Dragana, 2010, 2010 48th Annual Allerton Conference on Communication, Control, and Computing (Allerton), P302, DOI 10.1109/ALLERTON.2010.5706921
[6]  
Bajovic D., 2012, CONSENSUS PRODUCTS R
[7]   Sensor Selection for Event Detection in Wireless Sensor Networks [J].
Bajovic, Dragana ;
Sinopoli, Bruno ;
Xavier, Joao .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2011, 59 (10) :4938-4953
[8]   Distributed Detection via Gaussian Running Consensus: Large Deviations Asymptotic Analysis [J].
Bajovic, Dragana ;
Jakovetic, Dusan ;
Xavier, Joao ;
Sinopoli, Bruno ;
Moura, Jose M. F. .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2011, 59 (09) :4381-4396
[9]   Distributed detection with multiple sensors .2. Advanced topics [J].
Blum, RS ;
Kassam, SA ;
Poor, HV .
PROCEEDINGS OF THE IEEE, 1997, 85 (01) :64-79
[10]  
Boyd S., 2004, CONVEX OPTIMIZATION, VFirst, DOI DOI 10.1017/CBO9780511804441