Document Type : Original Article
Authors
1
1. Communication Department, Faculty of Engineering at Shoubra, Benha University, Cairo, 11629, Egypt. 2. Electrical Department, Faculty of Engineering, MTI University, Cairo, 11439, Egypt.
2
Communication Department, Faculty of Engineering at Shoubra, Benha University, Cairo, 11629, Egypt.
3
1. Communication Department, Faculty of Engineering at Shoubra, Benha University, Cairo, 11629, Egypt. 2. Faculty of Computer Science, Benha National University, Cairo, 13518, Egypt.
4
Communication Department, Faculty of Engineering at Shoubra, Benha University, Cairo, 11629, Egypt
Abstract
Abstract: Mobile Edge Computing (MEC) is vital for next-generation low-latency services, enabling resource-constrained mobile devices to offload intensive tasks to cloud-based infrastructure. This reduces energy consumption and latency, making MEC a key component of future mobile networks. Utilizing the Cloud-Radio Access Network (C-RAN) architecture, which integrates Baseband-Units (BBU) with MEC servers and Remote-Radio-Heads (RRHs), complex tasks are executed closer to users, improving service quality and creating new revenue streams for network operators. This paper examines computational offloading profitability from a network operator’s perspective. The offloading process involves optimizing radio and computational resources, posing a non-deterministic polynomial-time (NP) hard problem. To address this complexity, four optimization algorithms are evaluated: Ant-Colony Optimization (ACO), Normal-Genetic-Algorithm (NGA), Fast-Genetic-Algorithm (FGA), and Modified Spectrum Efficiency-Based Joint Optimization for Offloading and Resource-Allocation (Modified SJOORA). ACO, inspired by ant behavior, seeks optimal paths, while NGA and FGA simulate natural selection, with FGA offering faster convergence. Modified SJOORA is designed to optimize resource-allocation in MEC environments. The study compares these algorithms under various conditions to identify the most effective for profit maximization. Additionally, a novel approach reduces computational time by using a strategically seeded population and regression-based machine learning to estimate resource-allocation, maintaining accuracy while enhancing efficiency.
Keywords
)
View on SCiNiTO