Optimisation of Resource Allocation in High User-density Wireless Networks

摘要

Combined with technological advancements, resource allocation is a key tool used to improve the performance of wireless networks. Today's heterogeneous networks are mainly interference limited making resource allocation an important tool in interference management. Of particular interest in this thesis is the use of resource allocation for interference avoidance. In this thesis, a power dimension is added to established radio resource management (RRM) techniques to improve the throughput attained in small cell networks. Two power allocation techniques are proposed to this end. The first technique uses power allocation to modify the interference map generated in order to increase radio resource reuse and throughput. The second technique increases reuse after initial allocation where all base stations transmit the same power to their users. In the second allocation, some users are selectively granted access to specific resources if they can reduce their power to meet the interference constraints of incumbent users, allocated in the initial allocation. Simulation results for both power allocation techniques show that they attain increased throughput while maintaining the minimum data rate performance of the network. In most 802.11 networks, resource allocation is distributed and tends to be limited to channel selection and in some cases power allocation. However in 802.11ah networks, which are designed for wide coverage and thousands of sensor type devices, there is some centralisation in the form of restriction of which stations (STAs) are permitted to contend for access to the wireless medium at a given time. The contention works fairly well in low traffic scenarios without hidden nodes. However, in saturated scenarios with hidden nodes, its performance is poor. The research presented in this thesis proposes a grouping algorithm to solve the hidden node problem. Furthermore it proposes that backoff timers set during the contention process are set based on the unique association identifiers (AID) of the group members. This ensures that collisions due to devices choosing the same backoff timer value are eliminated. An analytical model is developed and is verified through simulations. Further simulations show an improvement in throughput, delay and energy efficiency when AID-based backoff timers are used compared to the standard random backoff timers in saturated scenarios. In a network using AID-based backoff and the grouping technique that manages hidden nodes, this research proposes a solution to the outage problem in 802.11ah networks which involves the use of connected stations (STAs) to relay packets for neighbouring STAs in outage. The results obtained through simulations show that for dense networks, using connected STAs as relays can help solve the outage problem, in the absence of relays.