Optimal Joint Allocation Of Wireless And Computational Resources For Mobile Edge Computing (MEC) Systems In Industrial Internet

Mobile edge computing(MEC)network refers to the wireless network where computing units are implemented in the vicinity of mobile users so that they can assist users to compute their tasks.With the rapid development of the Internet of Things(Io T)and the Industrial Internet,the MEC network has become a hot topic in wireless communications,the Io T network,and the Industrial Internet.Traditional cellular networks allocate three types of communication resources,namely,the base station(BS),transmission power and bandwidth,to users during each scheduling period.For MEC networks,there is one more type of resource,i.e.,the computing resource to be allocated.This thesis focuses on the joint allocation of communication and computational resources in MEC networks.The major contributions of this thesis is summarized as follows:First,in a single-cell OFDMA MEC network where data downloading users and computation offloading users coexist,the joint allocation problem of sub-channels,uplink and downlink transmission power,and computational resource is studied.The aim of this problem is to maximize the sum rate of downloading users and minimize the sum delay of computation offloading users.The corresponding optimization problem is first formulated.To obtain an effective low complexity algorithm of this non-convex NP-hard problem,the original problem is transformed into the iterative solution of three sub-problems: the sub-channel optimization sub-problem,the downlink transmission power optimization sub-problem,and the uplink transmission power and computational resource optimization sub-problem.The following methods are used to solve these three sub-problems: Firstly,the branchand-bound(Bn B)algorithm is used to solve the first sub-channel allocation sub-problem? Secondly,by analyzing the Karush-Kuhn-Tucker(KKT)conditions,the nonlinear equation satisfied by the optimal solution of the second sub-problem is obtained,based on which a bisection algorithm is designed?Thirdly,through analyzing the KKT conditions,the analytical expression of the optimal solution of the third sub-problem is obtained.Simulation results reveal the following conclusions: Firstly,the sum rate of downloading users increases with the maximum number of sub-channels for downloading users,while the sum delay of computation offloading users first increases and then remains constant?Secondly,the sum delay of computation offloading users and the sum rate of downloading users both increase with the balancing coefficient in the objective function.Furthermore,in multi-cell OFDMA MEC networks,the joint allocation of BSs,sub-channels,uplink transmission power,and computational resources is studied aimed at minimizing the sum energy consumption of users.The corresponding optimization problem is formulated.To obtain an effective low-complexity algorithm of this non-convex NP-hard problem,the original problem is converted into the iterative solution of two sub-problems: the transmission power and computational resource allocation sub-problem,and the BS and sub-channel allocation sub-problem.For the first sub-problem,the optimal computational resource is derived as a function of the optimal transmission power,simplifying the sub-problem into a pure power allocation problem.Then,a distributed algorithm is used to decompose the power allocation problem into each BS,and the nonlinear equation satisfied by the local optimal solution of the problem is obtained which can be solved by a bisection algorithm.For the second sub-problem,two algorithms are proposed.One is the modified-cuttingplane(MCP)algorithm that combines the Bn B method and the mixed integer linear programming?The other is the pivoting and sub-gradient(PS)algorithm that combines the pivoting technique and the sub-gradient method.The MCP algorithm can obtain the optimal solution with exponential-time complexity,while the PS algorithm can obtain the sub-optimal solution with polynomial-time complexity.Simulation results show that the performance of the PS algorithm is close to that of the MCP algorithm and better than the benchmark Tuyen algorithm.Additionally,for multi-cell cooperative OFDMA MEC networks where data downloading users and computation offloading users coexist,the joint optimization of BSs,sub-channels,uplink and downlink transmission power,computing resources,and server cooperation ratio is studied aimed at maximizing the sum downloading rate and minimizing the sum delay of computing offloading users.The corresponding optimization problem is formulated.To obtain an effective low-complexity algorithm,the original non-convex NP-hard problem is converted into the iterative solution of three sub-problems: the BS and sub-channel optimization sub-problem,the uplink and downlink transmission power optimization sub-problem,and the computing resource and cooperation ratio optimization sub-problem.For the first sub-problem,the Bn B algorithm is used to solve it.For the second subproblem,an MTRSCA algorithm is proposed which combines the trust region algorithm,the SCA algorithm,and the geometric programming method.For the third sub-problem,an HLGPA algorithm is proposed which combines the linear approximation of norms and geometric programming method.Simulation results reveal the following conclusions: The average rate of downloading users decreases with the proportion of downloading users.When downloading users can occupy at most one sub-channel,the average delay of computing users decreases with the proportion of downloading users.When downloading users can occupy at most two sub-channels,the average delay of computing users increases with the proportion of downloading users.The sum delay of computing users and the sum rate of downloading users both increase with the maximum number of sub-channels that downloading users can occupy.The sum delay of computing users and the sum rate of downloading users both increase with the balancing coefficient in the objective function.When the data amount is small,the server cooperation ratio first decreases and then remains constant with the offloading delay.When the data amount is large,the server cooperation ratio first remains constant and then decreases with the offloading delay.Finally,for TDMA Unmanned-Aerial-Vehicle(UAV)data collection networks,the joint optimization of UAV trajectory and sensor uploading power is studied aimed at minimizing the UAV flying and hovering energy.The corresponding optimization problem is formulated.To obtain an effective low-complexity algorithm,the original non-convex NP-hard problem is transformed into the iterative solution of two sub-problems: the sensor serving order and UAV hovering position optimization sub-problem,and the sensor transmission power optimization sub-problem.The first subproblem is formulated as a standard traveling salesman problem(TSP)and solved with the cuttingplane algorithm.For the second sub-problem,two algorithms are proposed.The first is the PSPSCA algorithm that combines the pattern search method and the SCA algorithm? The second is the AQSCA algorithm based on the deduced relationship between sensor uploading power and UAV hovering positions.The PSPSCA algorithm can obtain the optimal solution with exponential complexity,while the AQSCA algorithm can obtain the sub-optimal solution with polynomial complexity.Simulation results show that the performance of the AQSCA algorithm is close to that of the PSPSCA algorithm,while they both outperform benchmark algorithms.