Matthew Seligman

Storage routing for DTN congestion control

The Delay Tolerant Networking (DTN) architecture approaches the problem of reliable message delivery in intermittent networks using a store-and-forward approach where messages may remain stored for relatively long periods of time in persistent storage at DTN routers. Forwarded messages are removed from persistent storage only when transfer acknowledgment to another router or final recipient is received. Congestion in such networks takes the form of persistent storage exhaustion. Several solutions exist including slowing sources, using alternative routes, discarding traffic, or migrating messages to alternative storage locations. We propose storage routing (SR), a congestion management solution of the last form. SR employs nearby nodes with available storage to store data that would otherwise be lost given uncontrollable data sources (such as sensors). SR determines a collection of messages and neighbors to migrate them to using a set of locally scoped distributed algorithms, possibly incorporating loops that are known to be optimal for some DTN routing scenarios and decouples storage management from global DTN route selection. Simulations show up to a 500 per cent performance improvement using SR as compared with a comparable scenario lacking SR. Furthermore, we show a desirable parameter insensitivity to node storage capacity, neighborhood search radius, and message lifetime. Copyright

Epidemic routing with immunity in Delay Tolerant Networks

In this paper, we modify and extend epidemic routing used in intermittent networks such as delay tolerant networks (DTNs). In particular, we propose to include immunity-based information disseminated in the reverse direction once messages get delivered to their destination. There are many variants of epidemic routing that are intended to result in better resource utilization by reducing the number of copies or through the use of sophisticated forwarding policy. Our focus is to use the information of already delivered messages in an immunity-list that will prevent any future exchange of those messages. Through the use of this technique we expect that the percent of delivered messages at lower delays will be higher because of better buffer and network utilization. Our simulation shows statistically significant performance improvement both in delivery ratio and delay for immunity-based epidemic as compared to the basic epidemic protocol.

Delay tolerant network routing: Beyond epidemic routing

In this paper, we identify two distinct classes of routing algorithms for Delay or Disruption Tolerant Networks (DTN). The purpose of this classification is to clearly delineate the assumptions they work under and to facilitate mapping of applications to these algorithms. Algorithms based on opportunistic contact and some variant of epidemic routing use minimal topology knowledge and the most resources due to replication. The island-based algorithms find routes between connected islands and are closer to real applications such as tactical military networks. The general consensus is that there is no single routing solution that will minimize delay at the same maximizing throughput for DTNs. Majority of the algorithms assume non-standard and diverse scenarios which makes comparative evaluation difficult. Mapping applications to algorithms also poses a problem as many of the known applications do not match in scale the rigor of the proposed algorithms. Further efforts in standardization and verifiable evaluation using application context are the way forward.

Delay-tolerant network experiments on the meshtest wireless testbed

Delay Tolerant Networks (DTNs) are a class of networks in which a contemporaneous end-to-end path from source to destination generally does not exist. Such networks use on a store-carry-forward communication model which relies on the mobility of nodes to transfer data between geographically separated nodes. DTN researchers have relied heavily on simulation for evaluation, due to the difficulty and expense of running live experiments with real devices running real DTN implementations. MeshTest is a laboratory-based multi-hop wireless testbed that subjects real wireless nodes running real DTN implementations to reproducible mobile scenarios. It uses shielded enclosures and an RF matrix switch to dynamically control the attenuation experienced between pairs of nodes. The testbed is an ideal platform for DTN testing, offering convenient experimental control and data management. We have installed the DTN2 Reference Implementation on wireless nodes within the testbed, and in this paper, we report on a series of experiments based on the well-known Data MULE model. Specifically, we investigate the effects of buffer limitations on the data MULEs and sensors node, velocity of the data MULEs, and bundle generation size and rate. We report results on message delivery rate and latency for varying experimental parameters. We found that an encounter between nodes does not guarantee a successful data transfer. In our experience, the quality and duration of the link, contention, and load on the nodes all influence its performance.

Immunity-Based Epidemic Routing in Intermittent Networks

In this research, we propose to modify and extend epidemic routing used in intermittent networks. In particular, we propose to include immunity-based information disseminated in the reverse once messages get delivered to their destination. The goal is to design a more efficient routing protocol in terms of resource utilization. The idea is to analyze and evaluate the network performance using an immunity scheme in the context of epidemic routing and its variants. The reverse dissemination of such information requires minimal resources and the tradeoff in timely purging of delivered messages can be significant. We are using ns2 to implement a detailed simulation of the proposed immunity-based epidemic routing.

Routing for Data Delivery in Dynamic Networks

In this paper, we present a routing algorithm for a class of dynamic networks called the delay tolerant networks (DTNs). The proposed algorithm takes into account the quintessential DTN characteristic namely, intermittent link connectivity. Assuming a store and forward type of network transfers, our main objective in designing routing algorithms for such an environment is to maximize the number of delivered messages subject to storage constraints on intermediate nodes. We modify the simple breadth first search (BFS) algorithm to take into account link activation/deactivation and find the quickest route possible between source and destination nodes. We adopt a message drop policy at intermediate nodes to incorporate storage constraint into data delivery. We also introduce the idea of storage domain where a few connected nodes act as a single storage unit by sharing the aggregated storage capacity of the nodes in the domain. We evaluate the routing algorithm with and without storage domain in an extensive simulation for two types of network topologies-flat and layered. We implement the proposed routing algorithm in ns2 and present an extensive performance analysis using metrics such as delivery ratio, incomplete transfers with no routes and dropped messages. The most significant simulation result shows that routing with storage domain mitigates the storage bottleneck at a gateway node for a layered network topology. For instance, the delivery ratio for storage capacity of 10 with storage domain surpasses the delivery ratio for storage capacity of 20 without storage domain

A demonstration of the meshtest wireless testbed for delay-tolerant network research

MeshTest is a laboratory-based multi-hop wireless testbed that can subject real wireless nodes running real DTN implementations to reproducible mobile scenarios. It uses shielded enclosures and an RF matrix switch to dynamically control the attenuation experienced between pairs of nodes. The testbed is an ideal platform for DTN testing, offering convenient experimental control and data management. We have installed the DTN2 Reference Implementation on the testbed nodes, and and have been using it to run experiments. Our current experimental scenarios are similar to the well known Data MULE and Message Ferry models, but a large variety of other experimental scenarios are possible.

Routing in intermittent networks using storage domains

In this paper, we present a routing algorithm for a class of networks where a contemporaneous end-to-end path may not exist at the time of data transfer due to intermittent links. Several examples of such networks exist in the context of sensor networks, mobile ad hoc networks and delay tolerant networks. The proposed routing algorithms follow a priori routing similar to source routing. Link state changes are assumed to be known ahead of time, for instance, due to planned duty cycling resulting in scheduled connectivity. The basic idea behind the proposed routing algorithms is to modify the breadth first search (BFS) algorithm to take into account link state changes and find the quickest route between source and destination nodes. We introduce the idea of time-varying storage domains where all nodes connected for a length of time act as a single storage unit by sharing the aggregated storage capacity of the nodes. This will help situations where storage is a limited resource. We evaluate the routing algorithm with and without storage domain in an extensive simulation. The delay performance of the proposed algorithms is conceptually the same as flooding-based algorithms but without the penalty of multiple copies. More significantly, we show that the Quickest Storage Domain (Quickest SD) algorithm distributes the storage demand across many nodes in the network topology, enabling balanced load and higher network utilization. In fact, we show that for the same level of performance, we can actually cut the storage requirement in half using the Quickest SD algorithm. Copyright

Routing in intermittent network topologies

The topic of this paper is the algorithmic development of routing techniques for Delay Tolerant Networks (DTNs). Assuming a store and forward type of network transfers, our main objective in designing routing algorithms is to minimize the delay and maximize delivery subject to storage constraints on intermediate nodes connected by intermittent links. We present a novel modification to the breadth-first search algorithm to find the quickest route between a given source and any destination node in a delay-tolerant network. This is done without flooding the network - at any one time we maintain only one copy of the message in the network. We implement the proposed routing algorithm in NS2 and present an extensive performance analysis using metrics such as delivery ratio, incomplete transfers with no routes and dropped messages.