Delay Tolerant Mobile Social Networks Research Articles

Inter-Session Network Coding-Based Policies for Delay Tolerant Mobile Social Networks

We consider delay tolerant mobile social networks (DTMSNs), which are opportunistic networks made of human-carried wireless devices clustered into social communities. In such environments, routing is a challenge as the limited resources (such as memory and contact opportunities) must be efficiently used and shared between the sessions (or users, contents). To handle several unicast sessions, inter-session network coding (ISNC) has been proved necessary for optimal throughput in general networks, but is a delicate problem as it can quickly get detrimental. This paper investigates that ISNC can be beneficial to DTMSNs when used on top of a social-aware routing algorithm, whereas we exemplify and make explicit why any gain can hardly be expected with greedy replication, in regard to the current literature on ISNC. We then design decentralized criteria to control when and where in the network ISNC should be triggered, based on the node features (buffer size and social relationships) and network current load. These criteria are tested extensively on real-world contact traces, in terms of various metrics, such as number of deliveries, mean delay, or fairness. Our online ISNC protocol builds on the SimBet utility-routing policy. Our ISNC protocol can, however, run on top of any social-aware routing.

Social based throwbox placement schemes for large-scale mobile social delay tolerant networks

Mobile phone sensing is a new paradigm which takes advantage of smart phones to collect and analyze data at large scale but with a low cost. Supporting pervasive communications among mobile devices in such a large-scale mobile social network becomes a key challenge for this new mobile sensing system. One possible solution is allowing packet delivery among mobile devices via opportunistic communications during intermittent contacts. However, the lack of rich contact opportunities still causes poor delivery ratio and long delay, especially for large-scale networks. Deployment of additional stationary throwboxes can create a greater number of contact opportunities, thus improve the performance of routing. However, the locations of deployed throwboxes are critical to such improvement. In this paper, we investigate where to deploy throwboxes in a large-scale throwbox-assisted mobile social DTN. By leveraging the social properties discovered from real-life tracing data, we propose a set of social-based throwbox placement algorithms which smartly pick the location of each throwbox. Extensive simulations are conducted with a real-life wireless tracing dataset and a wide range of existing DTN routing methods. The results confirm the efficiency of the proposed methods.

Socially-Aware Multi-phase Opportunistic Routing for Distributed Mobile Social Networks

The work presented in this paper focuses on opportunistic routing in distributed Mobile Social Networks (MSNs). It proposes a novel routing protocol called Socially-Aware Multi-Phase Opportunistic (SAMPhO), where the routing procedure for each message is carried out in multiple phases. From the various socially aware metrics considered, ego-betweenness and tie strength are selected to be utilized according to the conditions of the social environment. Previous research on the area suggests that different metrics are useful in different conditions. However, only recently steps towards this direction are taken. To evaluate the proposed hypothesis, a simulated delay tolerant MSN was introduced in the OMNeT++ environment using a framework, developed as part of this work. It is called Socially-Aware Opportunistic Routing System (SAORS) and its modular design facilitates further research in opportunistic networks. The routing mechanism is divided in three independent stages, with SAORS focusing on providing the necessary functionality for the first one. The simulation results drawn clearly demonstrate the scalability and improved performance of SAMPhO, especially in highly social scenarios. It is shown to be both more efficient and effective than previous algorithms, in terms of bandwidth and memory utilization, and delivery rates respectively. However, it relies greatly on the accuracy of the social tie detection, since the copy generation is very limited.