Characterization and applications of temporal random walks on opportunistic networks

详细信息    查看全文

• 作者：Victor Ramiro ; ^a ; ^{victor.ramiro@isae-supaero.fr}Author Vitae ; Emmanuel Lochin^a ; ^{emmanuel.lochin@isae-supaero.fr}Author Vitae ; Patrick Sé ; nac^b ; ^{patrick.senac@enac.fr}Author Vitae
• 关键词：DTN ; Opportunistic network ; Temporal random walks ; Routing ; Monitoring
• 刊名：Computer Networks
• 出版年：2016
• 出版时间：24 December 2016
• 年：2016
• 卷：111
• 期：Complete
• 页码：29-44
• 全文大小：1373 K
• 卷排序：111

文摘

Opportunistic networks are a special case of DTNs that exploit systematically the mobility of nodes. When node contacts occur, routing protocols can exploit them to forward messages. In the absence of stable end-to-end paths, spatio-temporal paths are created spontaneously. Opportunistic networks are suitable for communications in pervasive environments that are saturated by other devices. The ability to self-organize using the local interactions among nodes, added to mobility, leads to a shift from legacy packet-based communications towards a message-based communication paradigm. Usually, routing is done by means of message replication in order to increase the probability of message delivery. Instead, we study the use of Temporal Random Walks (TRWs) on opportunistic networks as a simple method to deliver messages. TRW can adapt itself to the self-organizing evolution of opportunistic networks. A TRW can be seen as the passing of a token among nodes on the spatio-temporal paths. Since the token passing is an atomic operation, we can see it as forwarding one simple message among nodes. We study the drop ratio for message forwarding considering finite buffers. We then explore the idea of token-sharing as a routing mechanism. Instead of using contacts as mere opportunities to transfer messages, we use them to forward the token over time. The evolution of the token is ruled by the TRW process. Finally, we use the TRW to monitor opportunistic networks. We present the limits and convergence of monitoring the interact time between participating nodes.