Resource-efficient strategies for mobile ad-hoc networking

Abstract

The ubiquity and widespread availability of wireless mobile devices with ever increasing inter-connectivity (e. g. by means of Bluetooth, WiFi or UWB) have led to new and emerging next generation mobile communication paradigms, such as the Mobile Ad-hoc NETworks (MANETs). MANETs are differentiated from traditional mobile systems by their unique properties, e. g. unpredictable nodal location, unstable topology and multi-hop packet relay. The success of on-going research in communications involving MANETs has encouraged their applications in areas with stringent performance requirements such as the e-healthcare, e. g. to connect them with existing systems to deliver e-healthcare services anytime anywhere. However, given that the capacity of mobile devices is restricted by their resource constraints (e. g. computing power, energy supply and bandwidth), a fundamental challenge in MANETs is how to realize the crucial performance/Quality of Service (QoS) expectations of communications in a network of high dynamism without overusing the limited resources. A variety of networking technologies (e. g. routing, mobility estimation and connectivity prediction) have been developed to overcome the topological instability and unpredictability and to enable communications in MANETs with satisfactory performance or QoS. However, these technologies often feature a high consumption of power and/or bandwidth, which makes them unsuitable for resource constrained handheld or embedded mobile devices. In particular, existing strategies of routing and mobility characterization are shown to achieve fairly good performance but at the expense of excessive traffic overhead or energy consumption. For instance, existing hybrid routing protocols in dense MANETs are based in two-dimensional organizations that produce heavy proactive traffic. In sparse MANETs, existing packet delivery strategy often replicates too many copies of a packet for a QoS target. In addition, existing tools for measuring nodal mobility are based on either the GPS or GPS-free positioning systems, which incur intensive communications/computations that are costly for battery-powered terminals. There is a need to develop economical networking strategies (in terms of resource utilization) in delivering the desired performance/soft QoS targets. The main goal of this project is to develop new networking strategies (in particular, for routing and mobility characterization) that are efficient in terms of resource consumptions while being effective in realizing performance expectations for communication services (e. g. in the scenario of e-healthcare emergency) with critical QoS requirements in resource-constrained MANETs. The main contributions of the thesis are threefold: (1) In order to tackle the inefficient bandwidth utilization of hybrid service/routing discovery in dense MANETs, a novel "track-based" scheme is developed. The scheme deploys a one-dimensional track-like structure for hybrid routing and service discovery. In comparison with existing hybrid routing/service discovery protocols that are based on two-dimensional structures, the track-based scheme is more efficient in terms of traffic overhead (e. g. about 60% less in low mobility scenarios as shown in Fig. 3.4). Due to the way "provocative tracks" are established, the scheme has also the capability to adapt to the network traffic and mobility for a better performance. (2) To minimize the resource utilization of packet delivery in sparse MANETs where wireless links are intermittently connected, a store-and-forward based scheme, "adaptive multicopy routing", was developed for packet delivery in sparse mobile ad-hoc networks. Instead of relying on the source to control the delivery overhead as in the conventional multi-copy protocols, the scheme allows each intermediate node to independently decide whether to forward a packet according to the soft QoS target and local network conditions. Therefore, the scheme can adapt to varying networking situations that cannot be anticipated in conventional source-defined strategies and deliver packets for a specific QoS targets using minimum traffic overhead. ii (3) The important issue of mobility measurement that imposes heavy communication/computation burdens on a mobile is addressed with a set of resource-efficient "GPS-free" soluti ons, which provide mobility characterization with minimal resource utilization for ranging and signalling by making use of the information of the time-varying ranges between neighbouring mobile nodes (or groups of mobile nodes). The range-based solutions for mobility characterization consist of a new mobility metric for network-wide performance measurement, two velocity estimators for approximating the inter-node relative speeds, and a new scheme for characterizing the nodal mobility. The new metric and its variants are capable of capturing the mobility of a network as well as predicting the performance. The velocity estimators are used to measure the speed and orientation of a mobile relative to its neighbours, given the presence of a departing node. Based on the velocity estimators, the new scheme for mobility characterization is capable of characterizing the mobility of a node that are associated with topological stability, i. e. the node's speeds, orientations relative to its neighbouring nodes and its past epoch time. iii