Wireless Social Network. Design and management algorithms to improve reliability

Nowadays the most affordable means to bring wide band connectivity to rural areas is represented by wireless technology, which guarantees flexibility, scalability, ease of installation and low investments. Wireless technology in the unlicensed bands presents some disadvantages, when intended to cover large areas. Commercial operators indeed might deploy wireless networks in digital-divided areas exploiting licensed bands or hybrid configurations where the backhaul is realized by high performance licensed links. In the recent years, several Wireless Community Networks have been created and grown all over Europe, with some relevant cases represented by the deployments in Athens, Barcelona, Vienna. A Wireless Community Network is created by citizens interested to exchange knowledge, contacts and information on a transversal infrastructure, thanks to the use of wireless equipment working in the unlicensed frequency spectrum. As a matter of fact, such experiences must be started by skilled people, living in places characterized by dense urbanization. Wireless Community Networks could be regarded as a means to overcome the digital dived in countryside, but exporting a Wireless Community Network in rural areas is difficult because of lack of technical skills, and economic conditions not comparable to those found in a big city. In this contest, the research work has been dedicated to support the startup and consolidation of a social-based wireless network operator in a rural quasi-remote area. The idea is to realize network infrastructures capable to reduce the infrastructural digital divide, create technical skills and increase the digital culture of the population, with a low economic effort compared to what should be spent to commercial operators. Technologically speaking, the most efficient and affordable solution is represented by the exploitation of Wi-Fi technology. The latest version of the IEEE 802.11 standard offers exceptional possibilities in terms of throughput and latency, even if regulatory limitations and interferences may lower down the capabilities of the network, especially when it is intended to cover large areas. For this reason, a model for a managed deployment has been identified. The model has been realized following two major criteria. The first one is an efficient reuse of frequency channels and the protection from interferences in order to provide a good coverage of the territory taking into account the presence of self-generated and external noise. The second one is an efficient algorithm for channel access. Its goal is to avoid redundancy, minimize the latency and increase the throughput. This has been achieved through an hybrid system that accesses the channel with a combination of contention-free and contention-based protocols. The alternation between the two is based on the traffic load in the network. This model has been applied to a Wireless Community Network, that has been modified from an earlier version based in a single municipality (16 km2) where six base stations were mounting a total of 13 radios, to reach a dimension of 115 base stations with 346 radios, covering to 45 municipalities over 646 km2. The system alternates 11 channels offering consistent download and upload throughputs (12 Mbps and 8 Mbps respectively). This network is managed by an association for social promotion “Senza Fili Senza Confini”, the first non-profit Internet Service Provider in Italy, which nowadays counts 2842 members, without any saturation or bottleneck. Senza Fili Senza Confini is a successful project, where the growth of the members’ number has been statistically analyzed: it has been noticed that the subscription rate depends on the number of families of each village. Municipality counting less than 250 census families exhibit a 7% of the census families subscription at the first month and a constant subscription rate of 0.9% per month. Whereas, if the municipality counts more than 250 census families the subscriptions at first month are 4% of the census families and the average subscription rate is of 0.7% per month.