Various Throughput Enhancement Approaches in IEEE 802.11-based Wireless Networks

Abstract

Ever-increasing demands for high-speed communication systems and high spectral efficiency are driving the continued evolution of wireless communications technology. In parallel with this evolution, continuous efforts to improve the performance of wireless communication networks have been made, but many issues remain either unsolved or with plenty of room for improvement. In the study reported in this thesis, we conducted fundamental research on methods for enhancing the throughput from the bottom up by addressing the following issues in wireless communication networks. How do wireless stations quickly recognize each other autonomously? How do wireless stations adequately assign bits to each sub-channel and quickly adjust rates over all sub-channels for orthogonal frequency-division multiplexing (OFDM)-based wireless communications? How do wireless stations access and utilize a medium efficiently? How do wireless stations resolve collisions? Finally, what is the sensing slot duration that maximizes the achievable throughput for secondary users (SUs) under the constraint that the primary users (PUs) are sufficiently protected?First, we deal with the neighbor discovery problem in IEEE 802.11-based wireless networks. In the mobile network environment, wireless stations need to establish a link before data transmission begins, but links may be established intermittently due to the mobility of wireless stations or time-varying channel conditions. To establish a link, wireless stations have to discover neighbors within transmission range. Each wireless station has to broadcast a constant probing message to discover its neighbors promptly. However, most mobile stations are battery-powered and cannot afford energy-consuming persistent probing. It therefore seems reasonable to adopt periodic wake-up and rendezvous mechanisms in probing operations to save energy consumption. Given discovery latency, it is desirable for such opportunistic networks to discover neighbors while consuming a minimal amount of energy. Note that, in general, the higher the energy consumption, the lower is the discovery latency. In Chapter 2, we present a neighbor discovery single-channel protocol. We first describe a method for designing a rendezvous protocol for opportunistic networks. Next, we prove that the proposed protocol works well. Furthermore, a theoretical analysis demonstrates that the proposed protocol reduces active slots by about 30\% as compared to existing schemes. It may therefore save energy.Multi-carrier modulation has been adopted in many wireless and wired standards in the form of OFDM. In these standards, multi-carrier systems employ conventional multi-carrier modulation, which uses the same modulation scheme over all sub-carriers. However, the overall bit error rate (BER) or margin of these systems is dominated by the sub-carriers with the worst performance. In order to avoid performance degradation due to additional bit allocations when channel conditions are bad, adaptive bit allocation for multi-carrier systems is critical. In Chapter 3, we present a novel optimal algorithm with a target bit rate and fixed energy constraint. The proposed algorithm can approach near-optimal bit allocation with fewer computational steps by using a water-filling solution in the initial bit allocation step and then using a multiple-bit loading procedure to satisfy the target bit rate. The proposed algorithm was evaluated by comparing it with existing algorithms. Our numerical results demonstrate the computational efficiency and fast convergence of the proposed algorithm. To implement the proposed algorithm in a practical system, however, a sender-receiver pair has to exchange all the information on sub-carriers, a procedure that leads to large overheads. We hence provide a method for implementing the proposed algorithm, called rate adaptation (RA), over time-varying wireless carriers in such a way that overheads are significantly reduced.The distributed coordination function (DCF) is designed such that wireless stations can access a wireless medium, and has been used in a variety of applications as a wireless medium access control (MAC) protocol. However, packet collisions constitute a major issue that severely degrades wireless network performance, increases the related overheads, including back-off slots, inter-frame spaces, and physical (PHY) layer headers, and constrains the performance of the DCF. Considerable effort has been invested in reducing packet collisions and the associated overheads, but most of these have been only marginally successful in improving wireless network performance. In Chapter 4, we present a novel MAC protocol based on the IEEE 802.11 DCF. It incorporates a new function that allows stations to use implicit orderings for transmissions

consequently, the stations can avoid random access attempts. Implicit ordering without any control messages is assumed, which allows contention-free transmissions. We evaluate the performance of the proposed protocol using simulations, and we show that it outperforms legacy DCF, DCF with optimal contention window, and ideal DCF, with respect to channel utilization. Although legacy DCF adopts the carrier sense multiple access with collision avoidance (CSMA/CA) mechanism and a slotted binary exponential back-off (BEB) algorithm to reduce the subsequent collision probability, it has been shown that DCF still suffers from poor bandwidth utilization, and improvements in its performance are greatly constrained due to severe collisions. In Chapter 5, we propose a fast collision resolution scheme for OFDM-based wireless communications. Our approach uses a predetermined zero-energy symbol. Using this symbol, the proposed scheme makes it possible for a receiver to recognize which senders are transmitting packets simultaneously and to determine the retransmission order after collisions occur, which can facilitate an increase in channel efficiency. We show analytically that a number of practical challenges related to this approach can be easily resolved.Cognitive radio (CR) has attracted considerable attention as a supporting technology for addressing the problem of radio frequency shortages. In CR networks (CRNs), SUs are allowed to utilize the licensed spectrum band of PUs opportunistically when these bands are temporarily unused. Thus, SUs should monitor the licensed spectrum band to detect any PU signal before trying to access the bands. According to the sensing outcomes, SUs either should vacate the spectrum band or may use it to transmit. Hence, both the spectrum sensing accuracy and the overall throughput of SUs depend on the sensing time. That is, there is a fundamental tradeoff between the spectrum sensing time and the achievable throughput of SUs. To determine the optimal sensing time and improve the throughput of SUs, considerable effort has been expended in studies using the saturated traffic and ideal channel assumptions. However, these assumptions are no longer valid in practical CRNs. In Chapter 6, we provide the framework of an 802.11-based MAC for CRNs, and we analyze this framework to find an optimal spectrum sensing time under saturated and unsaturated traffic conditions. Through simulations, our analytic model is verified and the fundamental problem of the sensing-throughput tradeoff for CRNs is studied.