New Distributed Constraint Reasoning Algorithms for Load Balancing in Edge Computing

Edge computing is a paradigm for improving the performance of cloud computing systems by performing data processing at the edge of the network, closer to the users and sources of data. As data processing is traditionally done in large data centers, typically located far from their users, the edge computing paradigm will reduce the communication bottleneck between the user and the location of data processing, thereby improving overall performance. This becomes more important as the number of Internet-of-Things (IoT) devices and other mobile or embedded devices continues to increase. In this paper, we investigate the use of distributed constraint reasoning (DCR) techniques to model and solve the distributed load balancing problem in edge computing problems. Specifically, we (i) provide a mapping of the distributed load balancing problem in edge computing to a distributed constraint satisfaction and optimization problem; (ii) propose two DCR algorithms to solve such problems; and (iii) empirically evaluate our algorithms against a state-of-the-art DCR algorithm on random and scale-free networks.