Queue Stability-Oriented Communication And Computation Resources Allocation In Wireless Communication Systems

With the rapid development of various kinds of wireless services,there is a soaring demand for higher transmission rate and larger computation capacity.However,the communication and computation resources are rattier scarce in wireless communication systems,and hence we are faced with unprecedented challenges on how to efficiently utilize those limited resources to provide reliable and stable services for the users.Inability of satisfying the broadband and low latency requirements causes queue overflow and data loss,which further results in a sharp increase in error rate,a rapid decline in system stability,and even service outage.Therefore,the queue stability is essential in wireless communication systems,and it is of great significance to design protection mechanisms to make sure all the queues in the system are stable.This thesis guarantees queue stability in wireless communication systems from three aspects,namely,improving transmission capacity,reducing transmitting content and localizing the services,and further optimizes the delay performance.Specifically,1.Providing broadband services through spectrum aggregation.We establish a queuing mod-el for spectrum aggregation systems,and design joint channel and power allocation algorithms based on queue stability,and further analyse the average queue lengths in closed form.Also,some practical constraints are taken into consideration,e.g.,limited spectrum aggregation capabil-ity,circuit power consumption,queue estimation and channel estimation.In homogeneous spec-trum aggregation system,we develop an analytical framework for energy efficient scheduling for delay-constrained spectrum aggregation,and further obtain the trade-off between average power consumption and average queue backlog in closed form.In heterogeneous spectrum aggregation system,we establish a theoretic framework to study the fundamental benefit of aggregating hetero-geneous channels with system coexistence through interference management,and propose a mod-ified water-filling power management algorithm in closed form,which characterizes the influence of mutual interference,channel quality and queue lengthes on power allocation.The results of the analysis show that the average power consumption and the average delay trade-off in a linear way,and the transmission capacity can be significantly enhanced by embracing spectrum aggregation.2.Supporting massive high speed data transmission through cooperative multicast relay.We establish a two-hop cooperative multicast relay system,taking the transmission efficiency,fairness issue and queue stability into consideration,channel allocation and relay selection algorithms are designed accordingly.By combining multicast and scalable video coding,the amount of transmit-ting content can be greatly reduced.As for such multi-objection optimization problem,we design a lexicographically optimal relay selection and channel allocation algorithm,and further conduct an asymptotic equivalent analysis from a geometric perspective to prove the Pareto optimality and the unique property of the solution.To make sure the queue stability of the system,we establish an optimal low-complexity multicast relay selection scheme based on Lyapunov optimization and re-inforcement learning,and further analyse the influence of service rate distribution and the number of relays on performance.The results of the analysis show that cooperative multicast relay helps improving the system performance in multi-dimension through reducing the transmitting content and improving the transmitting condition.3.Offering computation and caching capability in close proximity to mobile devices through mobile edge computing and edge caching.To promote computing experience,we establish a dual queue system including a data queue subsystem and a computation queue subsystem,and develop an analytical framework for joint radio and computational resource allocation for multi-user mobile edge computing systems,and further obtain the upper bound of the average queue lengths.Under the context of limited cache space at both the user devices and the mobile edge computing servers,the steady-state analysis of the discrete-time controlled random walk process is conducted to derive the delay performance in closed form.To improve the quality of data services,we establish a dual queue system,and enforce a dynamic mapping between request queues and data queues such that nodes can retrieve the combined effect of content demands and network congestion from local queues only.Then,a distributed framework for joint data forwarding and caching is developed by embracing network utility maximization.The results of the analysis show that the capabilities of computation and communication can be significantly enhanced through mobile edge computing and edge caching,which further contribute to reducing over-the-air traffic and improving the delay performance.