Physical Layer Optimization for Wireless Sensing and Network Connectivity
- Author(s): Jiang, Feng
- Advisor(s): Swindlehurst, A. Lee
- et al.
Wireless sensor networks (WSNs) have been widely studied for detection and estimation problems. When a coherent multiple access channel is employed between the sensor nodes and fusion center (FC), each sensor takes a noisy measurement of the signal of interest, amplifies and forwards the measurement to a FC through a wireless fading channel, and the FC makes a decision about the presence of the signal and estimates its parameters based on the coherent sum of the signals from all the sensor nodes. To minimize estimation error or maximize probability of detection, the transmit power at the sensors is optimized under either sum or individual power constraints.
Most of the existing works assume that the FC is configured with a single antenna. It is well-known that multiple antennas can effectively increase the throughput of a wireless link, and in this thesis, we investigate how to exploit the benefit of the multiple antennas in WSN, and we study the detection and estimation performance of a coherent amplify-and-forward WSN, in which the sensor node has single antenna and the FC is configured with a massive number of antennas. When the perfect channel state information (CSI) is available at FC, we derive optimal closed-form sensor transmission gains to optimize the performance of Neyman-Pearson (NP) detector and the linear minimum mean-squared error estimator (LMMSE), and if CSI is unknown at FC, we find the optimal sensor transmission gains to maximize the deflection coefficient of the energy detector (ED). Regarding the energy efficiency, our analysis show that the performance of NP detector and LMMSE estimator remain asymptotically constant if the sensor transmit power decreases proportionally with the increase in the number of antennas, and for the ED which does not require CSI, we show that a constant deflection can be asymptotically achieved if the sensor transmit power scales as the inverse square root of the number of FC antennas.
Additionally, we consider the problem of optimize the sensor phase to minimize the estimation error at FC, when the FC has a limited number of antennas. Two phase optimization algorithms are proposed and the sensor selection problem is formulated and solved. In addition to the case with multi-antenna FC, we also investigate the optimal power allocation for the WSN with single-antenna FC, when the FC use sensor measurements as input for a Kalman filter to track a dynamic parameter of interest.
When a fixed network infrastructure is not available (e.g. in military or disaster response scenarios), we investigate how to use the multi-antenna unmanned aerial vehicles (UAVs) as a relay to improve the connectivity between the mobile sensor nodes and the FC, which may be separated by a distances greater than their communication range. Several algorithms are proposed to optimize the trajectory of UAV.