Research On Effective Capacity And Scheduling Algorithms In Random Access Networks

The essence of random access is distributed control,thus it has the capability to support the network scalability which is defined as the variability of the number of terminals in the network and the variability of the traffic carried by the network.So the random access has caught much attention in academia and industry.However,the published random access algorithms can only guarantee the priority-based QoS(quality of service),they cannot guarantee the delay-QoS.Recently,relying on the effective capacity theory,researches on statistical delay-QoS guarantees have gradually been hot.The effective capacity theory organically combines the queueing theory and the information theory,and it provides a wireless channel model in the link layer.Based on the effective capacity theory,some researchers studied resource allocation algorithms with statistical delay-QoS guarantees for the central scheduling systems.However,due to the mutual effect among service of terminals and the conflict between the random behavior and the delay-QoS guarantees,there has none work about the distributed random access with statistical delay-QoS guarantees.In this work,we devote to investigating the random access algorithms with the statistical delay-QoS guarantees based on the effective capacity theory.We focus on the wireless random access network,including the RF(radio frequency)random access and the optical wirelss random access.In this work,the scheduling algorithms for the random access networks are regarded as the random access algorithms.For the network where the effectiveness of carrier sensing is good,we adopt the random backoff and carrier sensing scheme to achieve distributed cooperation,and explore the ?-throughput optimal CSMA(carrier sense multiple access)algorithms with statistical delay-QoS guarantees.For the OWC(optical wireless communications)system where the effectiveness of carrier sensing is unsatisfactory,we adopt the MPR(multi-packet reception)techniques to eliminate collisions,and explore the algorithm for realizing the MPR as well as the adaptive MPR capability optimization algorithm.Further,we explore the ALOHA random access algorithms with statistical delay-QoS guarantees.The main contributions and innovative points of this work are as follows.(1)For the RF network with good effectiveness of carrier sensing,we propose the ?-throughput optimal CSMA-like random access algorithm,which both provisions the delay-QoS guarantees in a distributed mode and retains the throughput optimality.The concept of throughput optimality refers to achieving the maximum throughput under the condition of keeping the stability of the network,which is more significant for the real communication networks.We model the service of the CSMA random access network as CSMA Markov chain,and investigate the approach to separate out the service model of a link from the CSMA Markov chain.Then,we derive the effective capacity for each link;we propose a delay-QoS renewal scheme,and map the diverse delay-QoS requirements of links to the piecewise constraints on their effective capacity.Then we formulate a log-likelihood ratio maximization problem subject to multiple constraints on effective capacity.The optimum solution of the resultant optimization problem is achieved provided that the long-term link service rates converge to the corresponding arrival rates,as a result,the throughput optimality achieved under multiple delay-QoS constraints.We investigate the relation between the local queue length evolution and the recursion equations of Lagrange multipliers of the converted optimization problem for achieving global optimization locally.Finally,we obtain the relation between TA(transmission aggressiveness)and the local information such as queue length,renewed delay-QoS exponent,etc,and provide random access with statistical delay-QoS guarantees in a totally distributed mode.Additionally,we prove the ?-throughput-optimality of our proposed CSMA algorithm via queueing network stability analysis methods.Simulation results demonstrate the validity of our proposed CSMA algorithm.(2)We explore the MPR realization algorithm for improving the throughput for the star-topology OWC system.Considering the strong random feature of the channel in the OWC system,we adopt the RO-MMSE-SIC(reliability ordering with minimum mean square error successive interference cancellation)signal detection algorithm in PHY(physical)layer to make the coordinator of the OWC system have MPR capability.Then,we derive the BER(bit error rate)expression of the RO-MMSE-SIC signal detection algorithm with the consideration of propagation error in the SIC-type algorithm,and the derived BER expression is verified by Monte Carlo simulations.Based on the derived BER expression,we propose the adaptive MPR capability optimization algorithm so that the MPR capability is adapted to different channel situations for satisfying the BER requirement of the traffic.Simulation results show that the optimized MPR capability decreases with the increase of the randomness of the channel.(3)We study the ALOHA-like random access algorithms with statistical delay-QoS guarantees for indoor OWC system.Due to the directionality of light in the indoor OWC system,the effectiveness of carrier sensing is significantly limited.Thus we adopt the MPR techniques to weaken collisions for the OWC random access system.(1)For the OWC system with the common statistical delay-QoS requirement,considering the multiple stochastic characteristics in the random access OWC system including the stochastic blocked channel and the ALOHA mechanism,we derive the aggregate effective capacity of the MPR-aided OWC system based the effective capacity theory.Then we formulate the ALOHA-like random access with the common statistical delay-QoS guarantees as an aggregate effective capacity maximation problem.Finally,we employ the IWO(invasive weed optimization)algorithm to find the optimum access probability for each terminal.(2)For the OWC system with the heterogeneous statistical delay-QoS requirements,we adopt the SIC-type signal detection algorithm in the PHY layer for achieving MPR.The ALOHA random access in the MAC layer affects the probability of the set containing the simultaneous transmitting terminals,as a result,it affects the SINR(signal to interference plus noise ratio)of the SIC-type algorithm in the PHY layer;The SIC-type algorithm affects the achievable transmission rate of each terminal,as a result,it affects the statistical delay performance of a terminal and the throughput of the system.We propose the concept of feasible access state for the MPR-aided system,and analyses the probability distribution of the achievable transmission rate for each terminal which reflects the mutual effect between the SIC-type algorithm in the PHY layer and random access in the MAC layer.Then we derive the effective capacity for each terminal,and formulate the ALOHA-like random access with heterogeneous delay-QoS guarantees as a saturation throughput maximization problem subject to multiple statistical delay-QoS constraints.Finally,we employ the improved IWO algorithm based on the Pareto optimization to solve the optimization problem for obtaining the access probability of each terminal.The simulation results demonstrate that the MPR technique elimate collisions in the random access system,and our proposed algorithm fully utilizes the MPR capability of the system.