Research On Collaborative Computation Offloading Strategy Based On Fiber-Wireless Networks

With the rapid development of smart mobile device and mobile internet,the number of mobile users and the number of mobile applications are increasing at an unprecedented rate.In order to satisfy the people's requirement,the applications are emerging with the features of high-energy-consumption and computation-intensive.However,the mobile devices,even with the state-of-the-art technology,may not be able to meet the requirements of tasks in terms of time delay and energy consumption due to the limitation of physical size.Fortunately,mobile edge computing technology can overcome the inherent disadvantages of mobile devices to some extent by offloading computing tasks to the cloud server.But the existing researches on mobile edge computing offloading only take mobile edge cloud as the offloading point,and they ignore the central cloud with more abundant computing resources.With the development of passive optical network technology,the fiber-wireless(FiWi)network is envisioned to be a promising network architecture for supporting the centralized cloud computing(CCC)and mobile edge computing(MEC)simultaneously.Based on the fiber-wireless network architecture,in this thesis we leverage the computing resources of the central cloud and mobile edge cloud to study the collaborative computation offloading of computing tasks on mobile devices.For those tasks with high real-time requirements,an optimization problem with the goal of minimizing the total delay of computing tasks is proposed and formulized,under the constraints of limited number of wireless channels,mobile device energy consumption and computation task delay.The NP-hardness is proved by reducing it to the bin packing problem.To get a solution in polynomial time,two algorithms are proposed.One is a heuristic collaborative computation offloading algorithm(HCCOA),which compares the delay of tasks in different execution modes and prefers the one with the minimal delay as the offloading strategy.The other is a genetic algorithm(GA4CCO)for this optimization problem,which takes the solution obtained by random or by the heuristic algorithm as the initial solution,and gets a better one through the genetic operation.Numerical results corroborate that the total delay of computing tasks can be reduced by 4.34%and 18.41%via HCCOA and GA4CCO,respectively,in comparison with the existing algorithms.In addition,for the joint optimization problem of collaborative computation offloading,we also propose an optimization problem,aimed at minimize the weighted sum of energy consumption and delay.A greedy algorithm and a simulated annealing algorithm are designed to solve the special case with the same weight.Simulation results show that both algorithms can get effective solutions.