An Optimal probabilistic forwarding protocol in delay tolerant Networks

"An optimal probabilistic forwarding protocol in delay tolerant networks" is an article about dtn routing in mobihoc2009.ArticleThe author is Professor Jie Wu and one of his doctoral students. In this paper, a new route is proposed based on the optimal stop theory.Algorithm(Optimal probabilistic forwarding, OPF ).

In this article, we first propose a probability forwarding mode based on the number of hops limit. In this mode, each message records its own remaining hops. The limit on the number of hops also determines the number of messages that can be forwarded, so that the routing algorithm can be highly scalable when the network size is large.

In this paper, the transmission probability is defined as the function Pi, D, K, TR about the current node, target node, number of remaining hops, and remaining survival time. In this way, the optimal forwarding rule can be described as follows: a message forwarding process is considered as a copy of the message updated on the sending node and the receiving node respectively. When node I encounters node J, whether node I should forward the message to J depends on whether the two new copies are updated to increase the overall transmission probability. That is to say, if Pi, D, K-1, Tr-1 and PJ, D, K-1, Tr-1 joint probability than Pi, D, K, Tr-1, then forward the message, otherwise, no action is taken. In this paper, PI, D, K, TR are calculated based on the optimal stop theory.

OPF assumes that the movement of nodes shows long-term regularity. Specifically, the encounter time and frequency between nodes can be predicted through historical data. Another basic assumption is that each node knows the average encounter interval between all the nodes {I, j} in the Network II, J (this may not be a suitable constraint ). The remaining TTL value of the node is an integer multiple of the U of a certain time unit. It is assumed that each node encounter at the beginning of the time slot. In addition, messages are not transmitted continuously in a single time slot, that is, messages cannot be transferred from I to J and then transmitted to K in the same time slot.

The optimal forwarding rule depends on the inequality.

1-(1-Pi, D, K-1, Tr-1) * (1-pj, D, K-1, Tr-1) ≥ Pi, D, K, Tr-1

Is it true. The Calculation of P is based on the Inverse normalization method, which is divided into the probability of successful transmission in each communication opportunity, so that the function about Tr and K is represented as the function of Tr-1 and K-1.

At last, this article discusses the algorithm expansion problem. It relaxed the conditions of known global network information to only some known network information, and considered the situation of broadcast.

Simulation results show that the message delivery rate of OPF is 20% higher than that of the current advanced proxy forwarding route, and the cost of the algorithm is that the number of copies and latency increase by 5%.

In this paper, the optimal stop theory is introduced into dtn routing for the first time, and a reverse recursive type based on hop limit and TTL is given. However, there is still a lack of accuracy in the calculation of transmission probability. In the actual network, the encounter interval between nodes in the global network may not be known. You also need to consider connection duration, bandwidth, and other factors when communicating with nodes.