Research On Computation Offloading In MEC Networks Using NOMA

In mobile edge computing(MEC)networks,the computation capabilities of MEC servers can be fully utilized to assist the mobile terminal with limited computing resources in offloading and computing.The application of non-orthogonal multiple access(NOMA)in MEC networks can significantly improve the spectrum efficiency,reduce the delay and energy consumption.Most of the existing works focus on the energy consumption and delay minimization,however the study on NOMA-MEC networks in terms of successful computation probability has been limited.The successful computation probability is defined as the event that the targeted tasks can be successfully executed within a given delay budget.It can be regarded as an important criterion to measure the Quality of Service(Qo S)of latency-sensitive services.Therefore,we firstly analyze the successful computation probability theoretically,and propose a new hybrid computation offloading scheme.Subsequently,the successful computation probability,time delay and energy consumption are jointly considered to improve the proposed computation offloading scheme.The main content and contributions are summarized as follows:1.We analyze and derive a closed-form expression for the successful computation probability in MEC networks.We propose a hybrid computation offloading scheme that can operate in one of the following modes,namely complete local computation,complete computation offloading and partial computation offloading.The application scenarios of the modes are introduced separately.The scheme we proposed consists of two phases.In task offloading phase,the mobile user performs downlink NOMA and offloads subtasks to the MEC servers.In task computation phase,the mobile user and MEC servers compute subtasks simultaneously.The optimization problem is modeled as a maximization of successful computation probability by optimizing time slot during task offloading and computing,offloading ratios and power allocation.The proposed problem is solved using theoretical analysis and the successive convex optimization(SCA).Compared with the existing schemes,numerical results show the proposed hybrid computation offloading scheme has the advantage of higher successful computation probability.2.We jointly consider the impact of the successful computation probability and energy consumption in NOMA-MEC network,improve the proposed hybrid computation offloading scheme.The total energy consumption consists of two parts,which are the energy consumption of offloading and computing at all nodes(MEC servers and mobile user).Problem is modeled as a minimization of energy consumption by optimizing offloading ratios and power allocation while satisfying the successful computation probability constraint.The exhaustive search algorithm and successive convex optimization are adopted to solve the proposed problem.Numerical results verify the rationality of the proposed scheme.