Graduation Year

2011

Document Type

Dissertation

Degree

Ph.D.

Degree Granting Department

Electrical Engineering

Major Professor

Ravi Sankar, Ph.D.

Committee Member

Richard D. Gitlin, Sc.D.

Committee Member

Sudeep Sarkar, Ph.D.

Committee Member

Kandethody M. Ramachandran, Ph.D.

Committee Member

In-ho Ra, Ph.D.

Keywords

directional medium access control, cooperative scheduling, Rayleigh fading, relay communication, polarization

Abstract

Ad hoc network is intrinsically autonomous and self-configuring network that does not require any dedicated centralized management. For specialized applications such as, military operations, search-and-rescue missions, security and surveillance, patient monitoring, hazardous material monitoring, 4G (4th Generation) coverage extension, and rural communication; ad hoc networks provide an intelligent, robust, flexible and costeffective solution for the wireless communication needs.

As in centralized wireless systems, ad hoc networks are also expected to support high data rates, low delays, and large node density in addition to many other QoS (Quality of Service) requirements. However, due to unique ad hoc network characteristics, spectrum scarcity, computational limit of current state-of-the-art technology, power consumption, and memory; meeting QoS requirements is very challenging in ad hoc networks. Studies have shown cross layer to be very effective in enhancing QoS performance under spectrum scarcity and other constraints.

In this dissertation, our main goal is to enhance performance (e.g., throughput, delay, scalability, fairness) by developing novel cross layer techniques in single-hop singlechannel general ad hoc networks. Our dissertation mainly consists of three main sections. In the first section, we identify major challenges intrinsic to ad hoc networks that affect QoS performance under spectrum constraint (i.e., single channel). In the later parts of the dissertation, we investigate and propose novel distributed techniques for ad hoc networks to tackle identified challenges. Different from our main goal, albeit closely related; in the first section we propose a conceptual cross layer frame work for interaction control and coordination. In this context, we identify various functional blocks, and show through simulations that global and local perturbations through parametric correlation can be used for performance optimization.

In the second section, we propose MAC (Medium Access Control) scheduling approaches for omni-directional antenna environment to enhance throughput, delay, scalability and fairness performance under channel fading conditions. First, we propose a novel cooperative ratio-based MAC scheduling scheme for finite horizon applications. In this scheduling scheme, each node cooperatively adapts access probability in every window based on its own and neighbors‘ backlogs and channel states to enhance throughput, scalability and fairness performance. Further, in the second section, we propose two novel relay based MAC scheduling protocols (termed as 2rcMAC and IrcMAC) that make use of relays for reliable transmission with enhanced throughput and delay performance. The proposed protocols make use of spatial diversity due to relay path(s) provided they offer higher data rates compared to the direct path. Simulation results confirm improved performance compared to existing relay based protocols.

In the third section, we make use of directional antenna technology to enhance spatial reuse and thus increase network throughput and scalability in ad hoc networks. In this section, we introduce problems that arise as a result of directional communication. We consider two such problems and propose techniques that consequently lead to throughput, delay and scalability enhancement. Specifically, we consider destination location and tracking problem as our first problem. We propose a novel neighbor discovery DMAC (Directional MAC) protocol that probabilistically searches for the destination based on elapsed time, distance, average velocity and beam-width. Results confirm improved performance compared to commonly used random sector and last sector based directional MAC protocols. Further, we identify RTS/CTS collisions as our second problem which leads to appreciable throughput degradation in ad hoc networks. In this respect, we investigate and propose a fully distributed asynchronous polarization based DMAC protocol. In this protocol, each node senses its neighborhood on both linear polarization channels and adapts polarization to enhance throughput and scalability. Throughput and delay comparisons against the basic DMAC protocol clearly show throughput, scalability and delay improvements.

Share

COinS