A Study on Efficient Approaches for Distributing Workloads Effectively in Edge Computing Systems

For the benefits of cloud computing, many enterprise companies have moved their services and apps to the cloud. The centralized cloud architecture experiences high workload, congestion, and delay bottlenecks resulting in high amounts of data and rapidly growing digits of connected devices that consume cloud services. Edge Computing (EC) is consequently presented as a new paradigm to increase cloud capabilities close to the end devices. Here, the task allocation is mentioned as the workload distribution amid innumerable nodes in an edge computing network. Major difficulties in workload distribution include locating each task optimally based on its needs for computing capacity, storing data, and bandwidth of the network, and adjusting to network’s continuously changing nature. Algorithms for workload allocation can be centralized, decentralized, hybrid, or based on machine learning. The selection of technique relies on the particular application’s pre-requisites. Each approach has advantages and disadvantages. In greater detail, the choice of the best work distribution techniques depends on the configuration and architecture of the edge computing system, namely MEC, joint computing of edge, fog and cloud, P2P EC and much more. As a result, allotting the tasks in edge computing is an intricate, varied, as well as a difficult challenge which calls for delicate balancing act amidst multiple potentially competing goals, inclusive of resource-aware, energy efficiency, machine learning with latency, safety and quality of Experience (QoE). Recent years have seen a rise in the amount of research studies on edge devices’ work allocation optimization and performance evaluation. This paper compares and contrasts several methods for work load distribution, algorithms which are much optimized, and the communication network types which are often employed in edge computing systems.