Enhancing Graph Routing Algorithm of Industrial Wireless Sensor Networks Using the Covariance-Matrix Adaptation Evolution Strategy

Abstract

The emergence of the Industrial Internet of Things (IIoT) has accelerated the adoption of Industrial Wireless Sensor Networks (IWSNs) for numerous applications. Effective communication in such applications requires reduced end-to-end transmission time, balanced energy consumption and increased communication reliability. Graph routing, the main routing method in IWSNs, has a significant impact on achieving effective communication in terms of satisfying these requirements. Graph routing algorithms involve applying the first-path available approach and using path redundancy to transmit data packets from a source sensor node to the gateway. However, this approach can affect end-to-end transmission time by creating conflicts among transmissions involving a common sensor node and promoting imbalanced energy consumption due to centralised management. The characteristics and requirements of these networks encounter further complications due to the need to find the best path on the basis of the requirements of IWSNs to overcome these challenges rather than using the available first-path. Such a requirement affects the network performance and prolongs the network lifetime. To address this problem, we adopt a Covariance-Matrix Adaptation Evolution Strategy (CMA-ES) to create and select the graph paths. Firstly, this article proposes three best single-objective graph routing paths according to the IWSN requirements that this research focused on. The sensor nodes select best paths based on three objective functions of CMA-ES: the best Path based on Distance (PODis), the best Path based on residual Energy (POEng) and the best Path based on End-to-End transmission time (POE2E). Secondly, to enhance energy consumption balance and achieve a balance among IWSN requirements, we adapt the CMA-ES to select the best path with multiple-objectives, otherwise known as the Best Path of Graph Routing with a CMA-ES (BPGR-ES). A simulation using MATALB with different configurations and parameters is applied to evaluate the enhanced graph routing algorithms. Furthermore, the performance of PODis, POEng, POE2E and BPGR-ES is compared with existing state-of-the-art graph routing algorithms. The simulation results reveal that the BPGR-ES algorithm achieved 87.53% more balanced energy consumption among sensor nodes in the network compared to other algorithms, and the delivery of data packets of BPGR-ES reached 99.86%, indicating more reliable communication.

Evaluation of Graph Routing Single Objective Paths Using Pre-set Unequal Clustering

“…Fig. 3 demonstrates that the best single-objective paths of the GR algorithm with a pre-set unequal clustering topology had a higher PDR and a lower PMR than the mesh topology in [1], which indicated that more data packets reached the . The proposed method facilitated a decrease in the ratio of missing data packets and an improvement in the communication reliability of the entire network.…”

“…The deficiencies of using single-objective paths for Graph Routing in an IWSN mesh topology were presented in [1]: unbalanced energy consumption leading to hotspots; low coverage; short lifetime of the network among others. To address these challenges, the proposed method principally evaluates best single-objective paths using pre-set unequal clustering.…”

“…In our work [1], which is the first of its kind to use optimisation techniques in IWSNs, the GR paths of the mesh topology were selected using the CMA-ES in order to enhance the GR algorithm of the IWSNs according to the requirements of IWSN monitoring systems. These requirements consist of low end-to-end transmission time, balanced energy consumption, and high communication reliability.…”

“…Simulations were conducted using MATLAB R2021b on a Windows 10 workstation running on an Intel(R) core™ i7 processor with 16 GB RAM. The simulation parameters are presented in the table below and are similar to those used in our prior work [1], [5]. 50 or 100 sensor nodes were employed in order to verify the performance of the POE2E, POEng and PODis, respectively, under varying node densities and random deployment.…”

“…As mentioned, we proposed three paths with a singleobjective function using the Covariance-Matrix Adaptation Evolution Strategy (CMA-ES) [1] based on the requirements of IWSNs. These are: the path based on the minimum distance between sensor nodes in the direction of the (PODis); the path based on maximum residual energy (POEng); and the path based on minimum end-to-end transmission time (POE2E).…”