Research On Dynamic Resource Allocation And Scheduling Design For High-Speed Railway Mobile Communications

Railway is the backbone of national integrated transportation system, acting as the national critical infrastructure, the main artery of the national economy and popular ve-hicles. In recent years, high-speed rail has been developed rapidly around the world, and China has constructed the longest high-speed rail. High-speed rail is an important carrier of China’s "Go Globe" strategy, and large-scale high-speed rail network will have a fun-damental impact on the existing transportation pattern, and meanwhile have an important strategic significance for the national economy and modernization.Mobile communication systems for high-speed rail act as nerve centers to guarantee the safe and effective operation of high-speed trains, and especially play a crucial role in providing data service transmission between the train and the ground. Future high-speed rail mobile communication systems not only provide control and monitoring services for the safe operation of high-speed trains, but also offer multimedia services for the passengers with different requirements. In order to guarantee the efficient, reliable and real-time service transmissions as well as improve the system performance, the resource management methods for high-speed rail mobile communications should be well studied.Taking full considerations of the dynamic characteristics and heterogenous service features in high-speed railway communication scenarios, this dissertation tries to jointly utilize mathematical modeling, theoretical analysis and system simulation to investigate the dynamic resource allocation and scheduling problem in high-speed rail mobile com-munications. The main innovations of this thesis are listed as follows:1) In terms of the heterogonous service requirements and time-varying wireless channels, we investigate the utility-based resource management problem in high-speed railway mobile communications, jointly optimizing the power allocation along the time and the packet allocation among the services. The problem is formulated as a mixed-integer non-linear programming problem, and through the integer constraint relax-ation and problem transformation, the problem is solved based on bisection search method and greedy algorithm. Theoretical analysis and simulation results show that the weighted proportionally fair utility function can balance the power efficiency and rate fairness along the time, and achieve that both the power allocation and resource allocation for each service along the time are proportionally fair.2) By considering the reliability requirement and dynamic characteristics, the dynamic resource management problem in high-speed railway mobile communications with distributed network architecture is investigated, which jointly optimizes the admis-sion control, power control and resource allocation. A dynamic resource management framework based on cross-layer design is developed, and the dynamic resource man-agement problem is formulated as a stochastic network optimization problem. By applying the Lyapunov optimization theory, a dynamic resource management algo-rithm is proposed based on content controller cooperation and distributed computing. Both theoretical analysis and numerical simulations show that the proposed algorithm can achieve the tradeoff between the obtained utility and queue backlog, and it has low complexity and high robustness.3) Under the mobile relay-based high-speed railway communication networks, we inves-tigate the downlink two-hop service scheduling problem by taking into account time-varying wireless channels, stochastic packet arrivals and service delay requirements. The scheduling problem is formulated as a constrained Markov decision process, where the reduced state optimality equation and the value function approximation are used to reduce the problem dimension and address the state space explosion problem in Markov decision model. Based on the stochastic learning theory and subgradient projection method, a distributed online scheduling algorithm and the corresponding implementation mechanism are proposed. Numerical experiments are carried out to show the convergence of the proposed algorithm, and its performance improvement in terms of average delay and service delivery performances.4) In terms of more practical channel capacity formula for finite blocklength codes, an energy-efficient packet scheduling problem with reliability and delay requirements is investigated, which jointly optimizes code blocklength and transmit power. By ap-plying the implicit function theorem, we prove that the transmission energy is mono-tonically decreasing and convex with the code blocklength under some mild condi-tion. Moreover, we develop iterative algorithms based on recursive search method and successive upper-bound minimization method respectively for the convex and non-convex offline problems, as well as a rolling-window based online scheduling algorithm. Simulation results demonstrate that the traditional designs using the Shan-non capacity formula significantly underestimate the transmission energy for reliable communications, and further illustrate the importance of the considered problem.