Optimal interference mitigation with deep learningbased channel access in wireless body area networks

Wireless body area networks (WBANs) are essential for medical applications, especially in remote health monitoring, as they transmit crucial and time-sensitive data collected by nodes positioned around or within the body. However, the coexistence of WBANs with wireless channels can degrade performance due to interference. This study introduces OIM-DLCAM, an optimal interference mitigation scheme for WBANs, which utilizes a deep learning-based channel access method. OIM-DLCAM addresses interference through the multiobjective Hungarian optimization (MOHO) algorithm, considering design constraints such as node transmission power, packet delivery ratio, and interference range. Additionally, it employs a deep probabilistic neural network-based channel access method (DPNN-CAM) to effectively mitigate interference by making decisions regarding contention window size, frame length, and buffer size. The proposed OIM-DLCAM scheme ensures fairness between users while enhancing system performance. Simulation results from both static and dynamic sensor node scenarios demonstrate its effectiveness under various conditions, showcasing its potential to improve WBAN performance in medical applications. The simulations reveal that OIM-DLCAM outperforms existing state-of-the-art schemes across various scenarios, with efficiency gains of up to 86.187%, 72.452%, and 47.954% for WBAN node density, mobility, and packet arrival rate, respectively. Moreover, it significantly reduces the average end-to-end delay and packet drop rate while improving throughput and packet delivery ratio compared with existing schemes. Additionally, comparisons with industry standards, such as the IEEE 802.15.4e norm, validate the suitability of OIM-DLCAM for cofounded WBANs.