On the effect of FTDMA techniques on a distributed estimation in WSN

This paper discusses an energy constrained distributed estimation in wireless sensor network (WSN) based on frequency and time division multiple access (FTDMA). In a measurement with WSN, sensors transmit observed quantized data to a fusion center (FC) via radio communication channels, to obtain a final estimate. Such systems require a multiple access technique such as time division multiple access, frequency division multiple access and FTDMA. In this paper, we focus on FTDMA as the multiple access technique, because of its high reliability and high spectral efficiency. We define an optimization problem to minimize a mean-squared error (MSE) while keeping a limited total power in FTDMA-based WSN, and derive the equations for bit allocation. Next, we investigate the effect of FTDMA-based measurement environment with WSN on the distributed estimation. Also, we show that a time slot and frequency band to be allocated to each sensor shall be optimized for the MSE minimization, because of the definition of the allocated power for sensors and the MSE. For this problem, we define a combination optimization to determine the time slot and frequency band. We develop the algorithm to determine the time slot, frequency band and allocated power to each sensor. In the algorithm, a genetic algorithm is employed to solve the combination optimization problem. Numerical examples are presented. Our developed algorithm outperforms conventional sensor selection with no optimization for the combination of sensors in terms of the MSE.