Heterogeneous Wireless Networks: An Analysis of Network and Service Level Diversity

Thumbnail Image


umi-umd-1412.pdf (661.44 KB)
No. of downloads: 2216

Publication or External Link






Future wireless systems will be a collection of symbiotic and hierarchical networks that address different aspects of communication needs. This architectural heterogeneity constitutes a network level diversity, where wireless domains can benefit from each other's spare resources in terms of bandwidth and energy. The dissertation investigates the network diversity through particularly interesting scenarios that involve capacity-limited multi-hop ad hoc networks and high-bandwidth wired or wireless infrastructures.

Heterogeneity and infrastructures not only exist at the level of networking technologies and architectures, but also at the level of available services in each network domain. Efficient discovery of services across the domains and allocation of service points to individual users are beneficial for facilitating the actual communication, supplying survivable services, and better utilizing the network resources. These concepts together define the service level diversity, which is the second topic studied in our dissertation.

In this dissertation, we first focus on a large-scale hybrid network, where a relatively resource abundant infrastructure network overlays a multi-hop wireless network. Using a random geometric random graph model and defining appropriate connectivity constraints, we derive the overall transport capacity of this hybrid network.

In the sequel, we dwell upon hybrid networks with arbitrary size and topology. We develop a Quality of Service (QoS) based framework to utilize the joint resources of the ad hoc and infrastructure tier with minimal power exposure on other symbiotic networks that operate over the same radio frequency bands. The framework requires a cross-layer approach to adequately satisfy the system objectives and individual user demands. Since the problem is proven to be intractable, we explore sub-optimal but efficient algorithms to solve it by relying on derived performance bounds.

In the last part of the dissertation, we shift our attention from network level diversity to service level diversity. After investigating possible resource discovery mechanisms in conjunction with their applicability to multi-hop wireless environments, we present our own solution, namely Distributed Service Discovery Protocol (DSDP). DSDP enables a highly scalable, survivable, and fast resource discovery under a very dynamic network topology. It also provides the necessary architectural and signaling mechanisms to effectively implement resource allocation techniques.