Abstract
The main theme of this thesis is about enhancing the performance of the Multipath Transmission Control Protocol (MPTCP) in Internet of Things (IoT) wireless networks. MPTCP is a transport layer protocol that simultaneously transmits data through multiple paths using several network interfaces (e.g., Wi-Fi, 5G and Ethernet). This has attracted a lot of research in industry and academia. However, the next generation IoT applications are expected to have strict Quality of Service (QoS) requirements. Traditional MPTCP is not able to satisfy such requirements, therefore, there is an opportunity to improve the protocol.This thesis first proposes a cross-layer design to enhance MPTCP. A network utility maximisation framework is formulated with joint congestion control, rout- ing, and scheduling design for ad-hoc networks. The objective of the optimisation problem is to maximise user utility, subject to source rate, scheduling, and queuing delay constraints. This joint optimisation problem is solved using the Lagrangian and two scheduling algorithms, specifically perfect scheduling and distributed scheduling, are proposed. The stability and convergence of the above algorithms are proved in fixed channels and time-varying channels scenarios. The results reveal that the proposed multipath algorithms outperform the existing schemes in terms of improving source rate and moderating congestion price.
The thesis also compares several variants of MPTCP to determine the best protocol in wireless IoT networks environments. The candidates are conven- tional MPTCP, MPTCP-TSC (Traffic Split Control) and ReMP TCP (redundant MPTCP). ReMP TCP provided the lowest number of transmissions and the lowest delay. To further diminish delay, the thesis introduces a routing technique, called Opportunistic Routing (OR), to the above protocols. OR is a networking protocol in which the traffic is broadcast to all wireless nodes that can hear the transmission.
Hence, the reliability of correct data transmission in a network is increased. The simulation results show that all OR-based MPTCP schemes are superior to the schemes without OR. Also, the most efficient joint selection is OR with ReMP
TCP.
To reduce the latency of MPTCP further and improve its reliability, the thesis investigates the feasibility of applying Full Duplex (FD) technology to MPTCP in Internet of Vehicles (IoV) networks. FD-capable devices can transmit and receive on the same radio frequency bandwidth concurrently. This property is ex- ploited by the MPTCP protocol; and a new design, called the FD-based multi-path transmission control protocol (FDMP), is proposed for ultra-reliable low-latency communication (URLLC) applications. FDMP incorporates a modified sched- uler and congestion control mechanism, as well as a proactive acknowledgment (ACK) mechanism and a novel re-transmission strategy. The simulation results demonstrate that FDMP outperforms the benchmark MPTCP and FD is feasible.
| Date of Award | 1 Jan 2022 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Mohammed Shikh-Bahaei (Supervisor) & Mischa Dohler (Supervisor) |