Efficient Offloading Schemes For MEC Networks With NOMA

With the rapid development of mobile Internet technology and the 5th Generation Mobile Communications(5G)communication technology gradually put into use,emerging mobile applications have high requirements for latency and computing power,and the performance of mobile devices is facing great challenges.Mobile Edge Computing(MEC)sinks computing and storage resources to the vicinity of mobile devices,using the computing power of MEC servers to complete task calculations and return results to mobile devices.At the same time,the traditional Orthogonal Multiple Access(OMA)technology can no longer meet the connection needs of today’s massive devices,which restricts the development of MEC.Non-orthogonal Multiple Access(NOMA)provides a more efficient user access scheme to improve spectrum efficiency and reduce the time and energy consumption of task transmission,and the application of NOMA technology to the transmission of user tasks in MEC systems can effectively improve the service experience of users.The existing NOMA-MEC network still faces many technical constraints: first,due to the difference in channel conditions between the users in the cell and the MEC server,users farther away from the MEC server need longer delay and higher energy consumption to offload their tasks,which causes a waste of communication resources and is detrimental to the transmission of other users’ tasks under the system delay constraint.The current research on NOMA-MEC systems focuses on the minimization of the system consumption delay and energy,while the successful computation probability is an important reference for the system task completion for delay-sensitive tasks.In addition,users face the problem of transmission interruption due to insufficient power of the device during task offloading.Existing research focuses on jointly optimizing computational and communication resources to reduce the energy consumption of the system,but it is difficult to meet the needs of more complex situations by resource allocation alone.Another solution is to introduce Wireless Power Transfer(WPT)technology to solve the energy constraint problem by supplying energy to user devices for task offloading and computing.However,cell edge users are susceptible to the "doubly near-far effect",which often increases the overall energy consumption of the system to ensure the quality of service for edge users.To address the above problems,this thesis designs efficient computation offloading schemes for NOMA-MEC networks,and conducts an in-depth study from two aspects:1.A NOMA-MEC network with user cooperation is designed,in which users closer to the MEC server utilize their own computing resources to assist remote users in computing tasks.The successful computation probability of the proposed cooperative NOMA-MEC network is analyzed theoretically,and closed-form expressions for the successful computation probability of the proposed scheme and the traditional NOMA/OMA uplink offloading scheme are derived.By comparing simulations,it is demonstrated that under the time delay constraint the proposed user cooperation scheme can achieve higher successful computation probability and better system performance than the traditional offloading scheme under the time delay constraint.2.A NOMA-based user cooperative WPT-MEC offloading scheme is proposed,which adopts user cooperation to eliminate the "doubly near-far effect" of the WPT-MEC network.The energy consumption of the system is minimized while ensuring efficient computation offloading for both near and far users,including the sum of the energy consumption of the two users’ local computation and offloading phases.It is formulated as an optimization problem under the system time delay constraint,and this problem is non-convex.The optimal solution of the problem is obtained by jointly optimizing the time slot,task amount,power,and Central Processing Unit(CPU)frequency allocations,combining mathematical relations and Lagrangian dual method.The simulation results show the advantages of the proposed scheme in terms of energy consumption compared to schemes without user cooperation and without NOMA transmission,and it is concluded that the proposed scheme is more suitable for systems that need to handle delay-sensitive tasks.