Using Evolving Graphs to evaluate DTN routing protocols Luciana Arantes 1 and Alfredo Goldman 2 and M´arcio Vin´ıcius dos Santos 2 1
2
Projet Regal – INRIA/LIP6 (Universit´e Pierre et Marie Curie, CNRS) 104, av. du President Kennedy, 75016, Paris, France
Instituto de Matem´atica e Estat´ıstica - Universidade de S˜ao Paulo (IME)-USP Rua do Mat˜ao 1010, CEP. 05508-090, S˜ao Paulo, SP, Brazil.
[email protected],
[email protected],
[email protected]
One of the main challenges of Delay-Tolerant Networks (DTNs) is on how to define effective routing protocols. Contrarily to traditional networks, which use static routing protocols and guarantee end-to-end paths between nodes of the network, connectivity is intermittent in DTNs. A routing path between two nodes is dynamically built over the time, i.e., a connection between two intermediate nodes of a routing path is not necessarily established beforehand. Hence, it may happen that at a given time or period there is no direct path between two given nodes, however further future temporary connections between some intermediary nodes may provide a path over time, also called journey, between those two nodes. Temporary connections between nodes, denoted contacts, is thus strongly exploited by DTN routing protocols for message delivery. A DTN routing protocol must carefully select journeys between nodes for performance’s sake. Usually routing takes place using a store and forward mechanism and depends on several parameters such as time of request, availability of connectivity, message size, network traffic, transmission delay, etc. Moreover, different metrics can be exploited for supporting routing decisions like the number of hops between the given nodes, the arrival time of the sent messages at the destination, or even the end-to-end delay to transmit a message between two nodes. Therefore, the aim of a routing protocol is to minimize the chosen metric. Several DTN routing protocols have been proposed in the literature either based on deterministic approach where information about connectivity and flows is available or stochastic one which may assume no previous provided connectivity data. Examples of these protocols are the epidemic one [5], oracle-based one [3], and probabilistic one [4]. Even in the case in which they are predictable, both the dynamics of DTNs and network disruptions make the choice of a routing protocol and its performance evaluation non trivial tasks. Hence, our proposal is to use a graph theoretic model, in particular the Evolving Graph (EG) theory [1], in order to provide a framework for evaluating least cost routing algorithms which exploit different metrics. Concisely, an EG is a time-step indexed sequence of subgraphs, where the subgraph at a given time-step corresponds to the network connectivity at the time interval indicated by the time-step value. The model allows for arbitrary changes between two subsequent time steps, with the possible creation and/or deletion of any number of vertices and edges [1]. The dynamics of the network is considered to be predictable and the routing algorithms are deterministic. The results obtained through the EG journeys will then be considered as optimal solutions and can thus be used as lower bounds for comparison with other existing proto-
cols like the ones listed above. In other words, optimal routing algorithms that minimize one or more metrics for DTN can be conceived using our EG framework. The current available algorithms for routing between two nodes on EGs are the shortest, fastest and foremost journeys. A similar work was done by Ferreira et. al. [2] which compares several well-known ad-hoc routing algorithms with the EG foremost routing algorithm. Evaluation performance experiments where conducted on top of NS2 simulator. In the DTN context we will run simulations on the ONE simulator 1 . In the figure below we can see some preliminary results comparing the algorithm performances on the deliverance probability when varying the communication range for the Helsinki scenario.
Figure 1. Comparison of Epidemic, Prophet and MaxProp routing schemes with the EG foremost and shortest algorithms
In [2] every node had full knowledge of the dynamics of the whole network which is in fact not realistic. Therefore, in order to propose algorithms which better reproduce the behavior of dynamic networks, we intend to consider a distributed approach where nodes have only partial knowledge of the network, i.e, we will assume that each node just has information about its current and future connectivity. We thus strongly believe that our proposal of an evolving graph framework will be an effective, useful and scalable tool for developing and comparing routing protocols for DTNs.
References [1] B. Bui Xuan, Afonso Ferreira, and Aubin Jarry. Computing shortest, fastest, and foremost journeys in dynamic networks. Internation Journal of Foundations of Computer Science, 14(2):267–286, 2003. [2] Afonso Ferreira, Alfredo Goldman, and Julian Monteiro. Performance evaluation of routing protocols for manets with known connectivity patterns using evolving graphs. Wireless Networks, accepted 2009, available on-line. [3] S. Jain, K. Fall, and R. Patra. Routing in a delay tolerant network. In SIGCOMM 04, pages 145–158, 2004. [4] A. Lindgren, A. Doria, and O. Scheln. Probabilistic routing in intermittently connected networks. In International Workshop on Service Assurance with Partial an Intermittent Resources, 2004. [5] A. Vahdat and D. Becker. Epidemic routing for partially connected ad hoc networks. Technical Report CS-200006, http://www.cs.duke.edu/ vahdat/ps/epidemic.pdf, 2000. 1
http://www.netlab.tkk.fi/tutkimus/dtn/theone/
