An efficient computation offloading in edge environment using genetic algorithm with directed search techniques for IoT applications

The contemporary computations in the IoT environment are often outsourced to remote infrastructures due to inherent computation-intensive nature of tasks or to circumvent the high expenditures associated with establishing local infrastructures. Traditional approaches, be it cloud-based or localized, are deemed impractical for computation offloading in this scenario due to the trade-off between benefits and increased latency which is unacceptable for many real-time IoT applications. Computation offloading at the edge environment is a promising solution in leveraging the untapped resources at Edge. Offloading application execution to edge servers offers the potential to reduce completion time and fulfill application requirements. However, the simultaneous offloading of multiple entire applications to edge servers may strain their hardware and communication channels and also poses significant challenge for ensuring Quality of Service. To overcome this, a multi-site workflow offloading problem is conceptualized wherein tasks are strategically distributed across various edge sites for efficient execution. The problem is combinatorial and an optimal solution is never guaranteed through deterministic procedures. Thus this work proposes a novel metaheuristic approach based on modified Genetic Algorithm with Directed Search (GADS) operators that aim at optimizing the overall Execution Time (ET) and Energy Consumed (EC). The efficiency of GADS is demonstrated against alternatives, and a relative comparison is carried out in terms of the quality of the solution and the pace of convergence to the solution and approximately a minimum of 1% and maximum 15% gain is achieved.