| Abstract |
Edge cloud computing enables computational tasks to be processed at the edge of the network using limited computational resources in comparison to larger remote data centres. Because of this, resource allocation and management is significantly more important. Existing resource allocation approaches usually assume that tasks have inelastic resource requirements (i.e., a fixed amount of bandwidth and computation requirements). However, this may result in inefficient resource usage due to unbalanced requirements from tasks that can cause bottlenecks. To address this, we propose a novel approach that takes advantage of the elastic nature of resources, recognising that the time taken for a task is generally proportional to the allocated resource. Therefore, we propose three mechanisms for solving this optimisation problem: an approximation algorithm and two auction-based mechanisms. Using extensive simulations, we show that considering the elasticity of resources leads to a gain in utility of around 20% compared to existing approaches, with the proposed approaches typically achieving 95% of the theoretical optimal. |