Performance Evaluation of Heuristic and AI-Based Load Balancing Techniques for Edge Computing Systems

Cloud Edge computing has acquire application areas like autonomous vehicles, smart cities, IOT etc. where latency is a significant factor. Traditionally cloud use to be a single point for processing and hence there was no closeness to the data source resulting in poor response time or latency and poor bandwidth usage while working on real time data. Although, the heterogeneous devices and internet availability in edge computing brings a lot of computing challenges in resource allocation and load balancing, required for high system performance, reliability, and energy efficiency. In this paper a performance evaluation traditional and AI-based load balancing techniques designed for edge computing systems is presented. Due to their simple implementation, minimal overhead, traditional heuristic-based load balancing algorithms such as Round Robin (RR), Least Connections (LC), and Weighted Least Connections (WLC) are popularly used. Such algorithms are best suited for predictable contexts as they function as per the pre-set rules or static criteria. Although, their accuracy in practical implementation is not so popular due to their inability to adjust according to the resource heterogeneity, user mobility, and workload fluctuations. Such methods expertise in situations where high fluctuations and can be optimized during resource allocation dynamically on the basis of past results or patterns available and energy constraints. Metrics like adaptability, scalability, fault tolerance, computational complexity, energy efficiency, and Quality of Service (QoS) delivery are analysed for both traditional & AI based methods in this study.