Target Tracking using Friendship Paradox

This paper considers the problem of target tracking using a network of social agents (ex. human operators). The target is assumed to evolve according to a finite state random process. A physical sensor, like radar, measures the underlying state of the target in noise, and the social agents form opinions on the state of the target using the information from the physical sensors. A controller aims to track the target by sampling the opinions of the social agents using two different sampling mechanisms: (i) Majority Opinion Sampling, and (ii) Friendship Paradox Sampling. The opinion sampling problem is formulated as a partially observed Markov decision process (POMDP). Our main results exploit the structure of the observation distributions of the controller, to construct myopic policies that provably upper bound the optimal policy of the sampling POMDP.