On the applicability of model predictive power control to an IEEE 802.15.4 wireless sensor network

Abstract

This paper addresses the practical applicability of dual mode infinite hori zon model predictive control (MPC) strategies for a variety of stationary and mobile transmission power control problems that arise naturally in wireless sensor network (WSN) systems. The key performance requirement is to maximise battery lifetime while also preserving sufficient Quality of Service (QoS) among all users. This is achieved by implementing a closed-loop MPC mechanism within a nominal state-space tracking er ror based dynamic model that uses the received signal strength related to the signal-to interference-plus-noise ratio (SINR) as a state feedback signal. The resulting controller is compared with a number of proven strategies and is experimentally validated using an IEEE 802.15.4 WSN test bed and across a number of different test scenarios.

Application of model predictive control in wireless sensor networks

“…Witheephanich and Hayes investigated the feasibility of adopting MPC in power control in a WSN [22]. They formulated the power control problem as in a closed-loop dual mode model predictive structure, where inner mode is used when the system is close to the desired operating point and outer mode used when the system far from the desired operating point [11].…”

“…where Ȗ, r, and Ș 0 are the signal to interference and noise ratio (SINR), RSSI and thermal noise, respectively, in logarithm scale; C is measurement of offset and the term 30 accounts for the conversion from dBm to dB [22]. MPC was successfully applied in the state-space form that resembles Equations (1) and (2), where the RSSI was treated as state vector, and the objective to steer the RSSI at a certain interval to provide QoS.…”

“…In [22], a linear relationship between received power and RSSI was assumed and the authors did not consider interference power components. Here, we future extended their model and proposed the following one for the received SINR…”

“…Then, we can use the quadratic optimization formulated in [22] to solve our MPC-based SINR tracking problem. In other words, at each step t, the control input u(0) is obtained through quadratic programming and applied to the system using (9).…”

Fram evaluation as unified memory for convex optimization algorithms

“…The choice for the algorithms to implement comes from their application in MPC, where the control problem requires solving Quadratic Programming (QP) in real time. The investigation of MPC for the FRAM sensor node is motivated by the recent studies in the applicability of MPC in the context of WSNs, where the technique has been used for power control to guarantee a specified Quality of Service (QoS) adapting the transmission power level which has to be minimized [8]. Furthermore, MPC has been adopted in WSN for other purposes such as resources' allocation and power management in energy harvesting [9].…”