Social Distance Aware Resource Allocation

Communication flows associated with human end-users will have an underlying social context which determines the importance of the communication. In other words, the network gives us the capability to communicate, but the reason to communicate is external to it. In order to facilitate communication flows, the network has to make decisions on resource allocation to the flows among several others (admission control, traffic policing and so on). If the network remains agnostic to the underlying social context of the communication flow, the resulting decisions will at best be sub-optimal when viewed along the social context dimension.;In this work, we measure the social context associated with a communication flow and incorporate it into resource allocation decisions at the network. We use the notion of social distance between end-users to measure corresponding context. We combine the social distance as declared by the end-user with the overall importance of the user in the social network to derive a social-network-wide social distance measure. Further, we define social distance aware utility functions by imposing maximum achievable utility bounds on the communication flows based on the social distance. This ensures that in an optimal allocation of resources, flows of the same traffic type get differentiated service based on the associated social distance.;We present the resultant resource allocations with respect to wireless networks. Specifically, we look at the case of voice flows competing for resources over an IEEE 802.11e QBSS, and provide theoretical as well as simulation results demonstrating that our social distance aware resource allocation (SDA) achieves higher network utility than IEEE 802.11e for every case considered. We also look at the case of both voice and video calls competing for resources and show that SDA achieves improved network utility as compared to IEEE 802.11e. The reason for this is that SDA allocates resources based on classifying flows through the social distance dimension, as compared to IEEE 802.11e which only takes into account the traffic type for classification.;When users are requesting content hosted on the network, the relationships between content can be used to determine relative importance of content. In the case of social content (Youtube), such relationships are well-defined and thus the social network is already determined. After defining a corresponding social distance measure for videos, we look at three centrality techniques (degree, closeness, betweenness) to determine which of the three performs optimally in determining the most accessed videos. Finally, we look at some of the applications we implemented to determine the feasibility of SDA in a campus environment.