Determining zone radius of zone routing protocol by employing recurrent neural network in the 5G wireless network

In addressing the imperatives of 5G wireless networks, the development of network infrastructure capable of accommodating connectivity demands across diverse innovative technologies while upholding high-quality network standards becomes utmost important. In this quest, the selection of an appropriate network architecture, given the need to foster dynamic, self-organizing networks. This study employs a wireless mesh network (WMN) as the foundational network infrastructure due to its adaptability in evolving environmental conditions. Furthermore, the incorporation of unlicensed spectrum harmonious with 5G New Radio, as outlined in 3GPP Release 16, is enacted. In the quest to formulate an efficacious routing protocol capable of exercising traffic control in the face of escalated mobile data utilization, it is essential to ensure Quality of Service (QoS) for end users while optimizing resource efficiency. The Zone Routing Protocol (ZRP) is adopted to cater to the multifarious challenges encountered by WMNs, encompassing the constant flux of network topology, power transmission intricacies, and asymmetrical connections. Notably, the efficacy of ZRP is intertwined with the zone radius parameter, necessitating a dependable approach to its determination. In Addressing this, the proposed approach introduces the long short-term memory recurrent neural network (LSTM-RNN) algorithm. This algorithm empowers ZRP to dynamically adjust the zone radius value in accordance with network performance metrics routing overhead, energy consumption, throughput, and user usage. A dataset comprising these input metrics is partitioned into training and testing subsets, aiding the algorithm in predicting the optimal zone radius value. The efficacy of this methodology is scrutinized in both static and mobile node environments. In terms of network capacity, a bandwidth of 300 Mbps, aligning with the requisites of 5G wireless network technology, is employed. The comparative evaluation of the proposed LSTM-RNN ZRP against conventional ZRP is conducted on the basis of network performance and performance measurement. The zone radius values derived for static nodes fall within the range of 2-6 for the proposed approach and 2-7 for conventional ZRP. Similarly, for both static and mobile node environments, the range of zone radius values spans 1-7 for both algorithms. Analysing the algorithm's performance metrics, including mean square error, error histogram, regression values, and time series response, affirms its effectiveness. Moreover, the network performance evaluation showcases distinctive trends. Notably, LSTM-RNN ZRP demonstrates enhanced throughput and diminished routing overhead and energy consumption compared to conventional ZRP, underlining its efficiency. The number of users reached by nodes is also higher with LSTM-RNN ZRP, elucidating its superiority in user engagement. The novelty of this research lies in the algorithm's operation within an unlicensed spectrum with a bandwidth capacity of 300 Mbps, congruent with the parameters of 5G New Radio in 3GPP Release 16.