A machine learning-based dynamic link power control in wearable sensing devices

The main research challenges on developing Wireless Body Area Networks (WBAN) are related to the quality of the communication link and energy consumption. This article combines a Transmission Power Control (TPC) with a packet scheduler mechanism to reduce packet loss and minimize energy waste and radio interferences during on-body communications. The proposed solution takes advantage of Neural Networks and Fuzzy Inference Systems for modelling nonlinear dynamical systems to describe the on-body channel as a function of the operating environment, the relative position of the user’’ arm, and the body posture. Simulations show improved system communication reliability (10% less data packet loss than communications carried out at 0 dBm) at the latency expense. Thus, this mechanism shows that if packets are transmitted uniquely at instants in which the radio channel quality is favourable for communications, these inherently unreliable links can become reliable.