Data Fusion Trees for Detection: Does Architecture Matter?

We consider the problem of decentralized detection in a network consisting of a large number of nodes arranged as a tree of bounded height, under the assumption of conditionally independent and identically distributed (i.i.d.) observations. We characterize the optimal error exponent under a Neyman-Pearson formulation. We show that the Type II error probability decays exponentially fast with the number of nodes, and the optimal error exponent is often the same as that corresponding to a parallel configuration. We provide sufficient, as well as necessary, conditions for this to happen. For those networks satisfying the sufficient conditions, we propose a simple strategy that nearly achieves the optimal error exponent, and in which all non-leaf nodes need only send 1-bit messages.

[1]  T. Cover.  Hypothesis Testing with Finite Statistics , 1969 .

[2]  Nils Sandell,et al.  Detection with Distributed Sensors , 1980, IEEE Transactions on Aerospace and Electronic Systems.

[3]  L. Ekchian,et al.  Detection networks , 1982, 1982 21st IEEE Conference on Decision and Control.

[4]  P.K. Varshney,et al.  Optimal Data Fusion in Multiple Sensor Detection Systems , 1986, IEEE Transactions on Aerospace and Electronic Systems.

[5]  L.W. Nolte,et al.  Design and Performance Comparison of Distributed Detection Networks , 1987, IEEE Transactions on Aerospace and Electronic Systems.

[6]  Ramanarayanan Viswanathan,et al.  Optimal serial distributed decision fusion , 1987, 26th IEEE Conference on Decision and Control.

[7]  John N. Tsitsiklis,et al.  Decentralized detection by a large number of sensors , 1988, Math. Control. Signals Syst..

[8]  J. Tsitsiklis,et al.  Explicit Solutions for Some Simple Decentralized Detection Problems , 1990, 1989 American Control Conference.

[9]  J. Tsitsiklis,et al.  Explicit Solutions for Some Simple Decentralized Detection Problems , 1989 .

[10]  Michael. Athans,et al.  On optimal distributed decision architectures in a hypothesis testing environment , 1992 .

[11]  D. Kleinman,et al.  Optimization of detection networks. I. Tandem structures , 1990, 1990 IEEE International Conference on Systems, Man, and Cybernetics Conference Proceedings.

[12]  E. Drakopoulos,et al.  Optimum multisensor fusion of correlated local decisions , 1991 .

[13]  John N. Tsitsiklis,et al.  Some properties of optimal thresholds in decentralized detection , 1992, Other Conferences.

[14]  Peter Willett,et al.  The suboptimality of randomized tests in distributed and quantized detection systems , 1992, IEEE Trans. Inf. Theory.

[15]  W. Gray,et al.  Optimal data fusion of correlated local decisions in multiple sensor detection systems , 1992 .

[16]  M. Athans,et al.  Distributed detection by a large team of sensors in tandem , 1992 .

[17]  Rick S. Blum,et al.  Optimum distributed detection of weak signals in dependent sensors , 1992, IEEE Trans. Inf. Theory.

[18]  D. Kleinman,et al.  Optimization of Detection Networks: Part 11-Tree Structures , 1993 .

[19]  J. Tsitsiklis.  Decentralized Detection' , 1993 .

[20]  John N. Tsitsiklis,et al.  Extremal properties of likelihood-ratio quantizers , 1993, IEEE Trans. Commun..

[21]  D. Kleinman,et al.  Optimization of detection networks with multiple event structures , 1994, IEEE Trans. Autom. Control..

[22]  Pramod K. Varshney,et al.  A unified approach to the design of decentralized detection systems , 1995 .

[23]  Rick S. Blum,et al.  Distributed detection with multiple sensors I. Advanced topics , 1997, Proc. IEEE.

[24]  Pramod K. Varshney,et al.  Distributed detection with multiple sensors I. Fundamentals , 1997, Proc. IEEE.

[25]  Amir Dembo,et al.  Large Deviations Techniques and Applications , 1998 .

[26]  Pramod K. Varshney,et al.  A Bayesian sampling approach to decision fusion using hierarchical models , 2002, IEEE Trans. Signal Process..

[27]  P.K. Varshney,et al.  Decision fusion rules in multi-hop wireless sensor networks , 2005, IEEE Transactions on Aerospace and Electronic Systems.

[28]  Wenjun Li,et al.  Distributed Detection in Large-Scale Sensor Networks with Correlated Sensor Observations , 2005 .

[29]  Venugopal V. Veeravalli,et al.  How Dense Should a Sensor Network Be for Detection With Correlated Observations? , 2006, IEEE Transactions on Information Theory.

[30]  Bin Liu,et al.  Channel-optimized quantizers for decentralized detection in sensor networks , 2006, IEEE Transactions on Information Theory.

[31]  Akshay Kashyap.  Comments on "On the optimality of the likelihood-ratio test for local sensor decision rules in the presence of nonideal channels" , 2006, IEEE Trans. Inf. Theory.

[32]  Moe Z. Win,et al.  On the Subexponential Decay of Detection Error Probabilities in Long Tandems , 2007, IEEE Transactions on Information Theory.

[33]  Peter Willett,et al.  On the optimality of likelihood ratio test for local sensor decision rules in the presence of non-ideal channels , 2003 .

[34]  Moe Z. Win,et al.  Bayesian Detection in Bounded Height Tree Networks , 2007, IEEE Transactions on Signal Processing.