Abstract:

To meet the requirements for secure and efficient computing offloading in edge computing, the computing offloading scheme should not only consider the cost optimization of normal computation task offloading, but also take into account the security costs arising from defending against DDoS attacks. In a multi-user, multi-task edge computing system, we proposed an edge-cloud collaborative computing offloading considering security (E-CCOCS). First, the computing offloading process was modeled, and a security model was used to detect malicious offloading traffic and generated trust values for all terminal devices based on the detection results. The joint offloading cost optimization problem of time delay, energy consumption, and security cost was formulated as a mixed-integer nonlinear programming problem. Second, for the cooperative offloading scenario of terminal devices-edge and edge-cloud, a double-layer improved particle swarm optimization algorithm (DLI-PSO) was proposed to solve the offloading cost optimization problem. Simulation results show that the convergence of the DLI-PSO algorithm is superior to the compared algorithms, and the cost of the E-CCOCS secure offloading scheme is lower than the cost of the compared schemes, effectively addressing DDoS attacks.

Key words: edge-cloud collaboration, computing offloading, security cost, particle swarm optimization, DDoS attack