An on-line sensor selection algorithm for sprt with multiple sensors

This paper presents an on-line sensor selection strategy (SSS) for the Sequential Probability Ratio Test (SPRT) with multiple sensors. We introduce an observation cost associated with every individual sensor, and aim to design an SSS that minimizes the expected total observation cost. The sensor selection rule is allowed to depend causally on the measurement values. Although the optimal SSS can be obtained by using methods such as value iteration, these methods are computationally quite demanding. In order to reduce the computational effort, we propose a new algorithm in which we partition the state space into three regions and solve for the SSS in each region. Numerical results show that the proposed algorithm can approximate very well the optimal SSS that minimizes the cost-to-go at every time step.