Interest-based Data Dissemination For Socially Aware Networking

The proliferation of smart devices enables mobile users (i.e. mobile nodes) to share diverse information. The large amout of information reflects the users’social properties. This causes the problem of how to utilize the social properties for better network performance and drives the emergence of socially aware networking (SAN). As a research area of SAN, data dissemination aims to studying mobile nodes’social properties to choose proper forwarders for network nodes. Howerver, most of the existing solutions do not explore further on how user interests and the relations between user intersts can affect the data dissemination.This thesis studies the relations between user interests, including interest inclusion, cross-layer interest, and interest intersection. Afterwards it proposes an interest and community based scheme (IntCom). IntCom presents a tree-based community structure for presentation of interest inclusion. The structure is used to update information of community density and social tie, provides evidence to choose proper forwarder. For performance evaluation, this thesis harnesses the ONE, devises simulations, compares IntCom with Epidemic and PROPHET with different values of buffer size, simulation duration and time to live (TTL), and tests IntCom’s performance over event interval and (α,β). The results show that IntCom achieves better performance than Epidemic and PROPHET in terms of delivery ratio, overhead and hop count, although the average latency is not the best. Furthermore, IntCom performs the best with10-20seconds’interval, while (α,β) pair affects IntCom slightly without a uniform trend.However, there are some problems about IntCom, i.e. IntCom only involves interest inclusion, and the mobility model is different from the real trace. Hence, this thesis introduces a geo-social distance based scheme (GSD). It supports interest intersection, and combines nodes’ geographical property and social information (i.e. interests) into the metric of geo-social distance for dissemination decision making. The experiment on GSD uses the dataset collected from ACM SIGCOMM2009to narrow the gap between mobility model and real trace. The experimental result shows GSD obtains the similar performance with IntCom, being the best in regard of delivery ratio, overhead and hop-count except for the average latency. But GSD remains the same performance with various values of event interval, ω, β, σ, τ and p_geo.