Sustainable Multi-MEC Task Offloading for 5G-Enabled Internet of Things Devices

Efficient task offloading strategies are essential in Multi-access Edge Computing (MEC) environments, particularly within fifth-generation (5G) networks, where inefficient management can lead to significant latency and increased rates of task drops. Given the rapid growth in connected devices and data-intensive applications, achieving scalability in MEC systems is vital for maintaining low latency, high reliability, and efficient data management. This study provides critical insights into the performance implications of varying the number of MEC servers, specifically examining dropped task ratios and latency. Utilizing Mixed Integer Linear Programming (MILP), we comprehensively assess how system performance scales with an increasing number of users and MEC servers. Our analysis demonstrates that MILP consistently yields superior results, effectively minimizing both latency and dropped task ratios even under substantial loads. In particular, scaling from 1 to 2 MEC servers yields a 57% reduction in the dropped task ratio, and further increasing from 2 to 4 MEC servers achieves an additional 53% reduction. Furthermore, in the scaling scenario of 1 to 2 MEC servers, MILP outperforms particle swarm optimization (PSO) by 33.3%, underscoring its effectiveness. In addition, our study investigates energy consumption by comparing two distinct scenarios: full offloading versus entirely local task processing. The analysis reveals significant energy savings through offloading, particularly for large image sizes, where offloading achieves up to a 52.44% reduction in energy consumption.