Abstract
The node buffer size has a large influence on the performance of Mobile Opportunistic Networks (MONs). This is mainly because each node should temporarily cache packets to deal with the intermittently connected links. In this paper, we study fundamental bounds on node buffer size below which the network system can not achieve the expected performance such as the transmission delay and packet delivery ratio. Given the condition that each link has the same probability p to be active in the next time slot when the link is inactive and q to be inactive when the link is active, there exists a critical value pc from a percolation perspective. If p > pc, the network is in the supercritical case, where we found that there is an achievable upper bound on the buffer size of nodes, independent of the inactive probability q. When p < pc, the network is in the subcritical case, and there exists a closed-form solution for buffer occupation, which is independent of the size of the network.
Highlights
The node buffer size has a large influence on the performance of Mobile Opportunistic Networks (MONs)
If p > pc, the network is in the supercritical case, where we found that there is an achievable upper bound on the buffer size of nodes, independent of the inactive probability q
We model the external factors in the network with the edge-Markovian dynamic graph (EMDG), which implies that: (1) States of each link vary from one time slot to another, and are i.i.d. among time slot t + 1 and the time slots before t. (2) The probabilities of being active or inactive are two constants p and q for all links, respectively
Summary
The node buffer size has a large influence on the performance of Mobile Opportunistic Networks (MONs). We study fundamental bounds on node buffer size below which the network system can not achieve the expected performance such as the transmission delay and packet delivery ratio. If p > pc, the network is in the supercritical case, where we found that there is an achievable upper bound on the buffer size of nodes, independent of the inactive probability q.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call