Deployment and operational aspects of rural broadband wireless access networks
Broadband speeds, Internet literacy and digital technologies have been steadily evolving over the last decade. Broadband infrastructure has become a key asset in today’s society, enabling innovation, driving economic efficiency and stimulating cultural inclusion. However, populations living in remote and rural communities are unable to take advantage of these trends. Globally, a significant part of the world population is still deprived of basic access to the Internet. Broadband Wireless Access (BWA) networks are regarded as a viable solution for providing Internet access to populations living in rural regions. In recent years, Wireless Internet Service Providers (WISPs) and community organizations around the world proved that rural BWA networks can be an effective strategy and a profitable business. This research began by deploying a BWA network testbed, which also provides Internet access to several remote communities in the harsh environment of the Scottish Highlands and Islands. The experience of deploying and operating this network pointed out three unresolved research challenges that need to be addressed to ease the path towards widespread deployment of rural BWA networks, thereby bridging the rural-urban broadband divide. Below, our research contributions are outlined with respect to these challenges. Firstly, an effective planning paradigm for deploying BWA networks is proposed: incremental planning. Incremental planning allows to anticipate return of investment and to overcome the limited network infrastructure (e.g., backhaul fibre links) in rural areas. I have developed a software tool called IncrEase and underlying network planning algorithms to consider a varied set of operational metrics to guide the operator in identifying the regions that would benefit the most from a network upgrade, automatically suggesting the best long-term strategy to the network administrator. Second, we recognize that rural and community networks present additional issues for network management. As the Internet uplink is often the most expensive part of the operational expenses for such deployments, it is desirable to minimize overhead for network management. Also, unreliable connectivity between the network operation centre and the network being managed can render traditional centralized management approaches ineffective. Finally, the number of skilled personnel available to maintain such networks is limited. I have developed a distributed network management platform called Stix for BWA networks, to make it easy to manage such networks for rural/community deployments and WISPs alike while keeping the network management infrastructure scalable and flexible. Our approach is based on the notions of goal-oriented and in-network management: administrators graphically specify network management activities as workflows, which are run in the network on a distributed set of agents that cooperate in executing those workflows and storing management information. The Stix system was implemented on low-cost and small form-factor embedded boards and shown to have a low memory footprint. Third, the research focus moves to the problem of assessing broadband coverage and quality in a given geographic region. The outcome is BSense, a flexible framework that combines data provided by ISPs with measurements gathered by distributed software agents. The result is a census (presented as maps and tables) of the coverage and quality of broadband connections available in the region of interest. Such information can be exploited by ISPs to drive their growth, and by regulators and policy makers to get the true picture of broadband availability in the region and make informed decisions. In exchange for installing the multi-platform measurement software (that runs in the background) on their computers, users can get statistics about their Internet connection and those in their neighbourhood. Finally, the lessons learned through this research are summarised. The outcome is a set of suggestions about how the deployment and operation of rural BWA networks, including our own testbed, can be made more efficient by using the proper tools. The software systems presented in this thesis have been evaluated in lab settings and in real networks, and are available as open-source software.