Sunday, March 4, 2012

Using Egocentric Networks for Content Distribution in Wireless DTNs

When we come across the term "Egocentric Network", we often start thinking of theoretical concepts like collaboration graphs and Erdos Numbers. I recently came across this very interesting practical application of Egocentric Networks in the area of wireless content distribution, which I would like to share with all of you.

The growing popularity of handheld devices, equipped with wireless interfaces like cellular network, Wifi and Bluetooth, along with the augment in their storage capacity, has resulted in a new communication paradigm - Delay Tolerant Content Distribution, where each individual carrying the device (node), can receive some content at a certain time, store it and then forward it to others at a later time.This store and forward mechanism can be made more efficient by using the concept of Egocentric Networks.

The nodes are ranked mainly based on their  Egocentric Betweenness. Other parameters like pattern of meeting between the nodes, connection quality, content accuracy (identifying the node which provides interesting data) are also taken into consideration. The Egocentric Betweenness Bi of a node i is defined as the number of pairs of neighbors of i that are not directly connected to each other.  Due to the dynamic nature of wireless Delay Tolerant Networks, Bi is averaged along time to give the Average Egocentric Betweenness. Nodes with higher value of the average egocentric betweenness are given higher ranks, as they have larger influence in the network because many nodes depend on them to make connections with other nodes.

The content exchange process works as follows - each node sends a content request to the highest rank node in its locality having content which is of interest to it. Locality here can be defined as the N-step neighborhood in the egocentric network of the node, for some empirical value of N. Let, for a particular case, the node requesting the content be A and the node receiving the request be B. B may receive multiple requests. If that happens, it accepts the request of the highest rank requesting node and rejects the others. Thus, the request from A can be accepted or rejected by B. If B accepts the request, A starts the download from B. The content is thus made available to A, which stores it and can forward it to other nodes on being requested. However, if B rejects the request, A sends the request to the next highest rank node and so on. Thus, the content is efficiently distributed among all nodes that are interested in it.

This type of distribution mechanism can be particularly useful for advertising, where the final objective is to achieve maximum dissemination, while taking the interest of the users into consideration.

[1] R. Cueves, E. Jaho, C. Guerrero, I. Stavrakakis: OnMove: A protocol for Content Distribution in Wireless Delay Tolerant Networks based on Social Information

No comments:

Post a Comment