TTL based routing in opportunistic networks

Abstract

One of the major problem in routing messages in an Opportunistic Network (OpNet) is the absence of a complete end-to-end path from the source to the destination. To deal with such adversity many routing protocols employ epidemic techniques to flood the network with multiple copies of a message. Although flooding many copies of a message in the network increases the chance of the message delivery, it can lead to a very high network overhead and can cause network congestion. On the other hand, the quota based routing protocols restrict the maximum number of copies of a message in the network by setting a replication factor. Such protocols maintain comparatively low overhead but suffer from low delivery ratio. In this paper, we introduce a novel quota based routing protocol, TTL Based Routing (TBR). TBR prioritizes both the schedule of messages to be forwarded to the neighbor and the schedule of messages to be dropped from the buffer. These priorities are based on massage time to live (TTL), message hop count, message replication count and message size. The TTL based message priority enhances the chance of message delivery by giving preference to the messages having the earliest deadline. Our simulation results show that, TBR achieves more than 10–15% higher delivery ratio than that of the available quota based routing protocols while incurring 10% to 15% less overhead. TBR not only matches the delivery ratio of flooding based routing protocols but also achieves better delivery ratio with 70% to 75% less overhead.

Introduction

Opportunistic Networks (OpNets) fall into a special category of wireless ad-hoc networks where a complete path from the source to the destination does not exist. In today's world, numerous opportunistic networking applications—such as Wildlife Tracking System (Juang et al., 2002, Jiang et al., 2009), Search and Rescue System (Huang and Cenwits, 2009), Underwater Sensor Network (Detweiller et al., 2009, Guo and Cui, 2010), Vehicular Ad-hoc Network (VANET) (Wang and Li, 2009, Harri and Filali, 2009), Pocket Switched Network (Chaintreau et al., 2005, Hui et al., 2005, Díaz et al., 2010), and people networks (Snowdon and Pollen, 2001, Liu and Wu, 2010), etc., are emerging. Disconnections and reconnections are pretty common in such networks. Factors such as high node mobility, low node density, intermittent power from energy management schemes, environmental interference and obstruction, short radio range and malicious attacks (Farrell and Cahill, 2006) etc. can ensue unstable connections in OpNets. In OpNets, the contacts are completely opportunistic, that is neither the meeting schedule nor the contact period of the nodes in the network is known in advance. Therefore, traditional Internet and Mobile Ad-hoc Network (MANET) routing techniques cannot be directly adopted in OpNets. This leads the research community to the development of specialized routing schemes that would better accommodate the characteristics of OpNets.

Generally, the routing algorithms in Wireless Ad-hoc Networks are stratified into two major categories—Forwarding based and Replication based. Forwarding based protocols use single copy of a message and attempt to direct the messages to the destination using the network dynamics. Traditional wireless routing protocols such as—AODV (Perkins and Royer, 1999), DSR (Johnson and Maltz, 1996), etc. are forwarding based protocols. Because of the unpredictability in network dynamics and irregularity in network connections forwarding based protocols are not well suited for OpNets. Still few forwarding based protocols (Spyropoulos et al., 2004, Burns et al., 2005, Henriksson et al., 2007, Liu and Wu, 2009, Guo and Cui, 2010) for opportunistic environments have been proposed. These protocols use the assumptions of network connectivity and environmental knowledge to take routing decisions. But their performance degrades when the environment becomes completely opportunistic.

Because of the inherent randomness in OpNets, most of the OpNet routing protocols employ replication based store and forward strategy. Replication based protocols insert multiple copies of a message into the network to increase the chance of message delivery. Most of the earlier replication based protocols tried to flood the messages to all the nodes in the network. Some examples of these Flooding based protocols are—Epidemic routing (Vahdat and Becker, 2000), PROPHET (Lindgren et al., 2003), MaxProp (Burgess et al., 2006), RAPID (Balasubramanian et al., 2007), PREP (Ramanathan et al., 2007), etc. Flooding based protocols generally replicate as many copies of a message as the resources permit. This approach of injecting many copies of a message may improve the delivery ratio but the approach is vulnerable to high network congestion.

To mitigate this flooding effect another variety of Replication based protocols named Quota based protocols were proposed. Quota based protocols keep the number of replicas of a message independent of the network size by setting up an upper limit on the maximum allowable replicas for a message. Spray and Wait (Spyropoulos et al., 2005), ORWAR (Sandulescu and Nadjm-Tehrani, 2008), Spray and Focus (Spyropoulos et al., 2007), and EBR (Nelson et al., 2009) are examples of some popular quota based protocols. Because of this limited flooding, quota based routing protocols usually fail to achieve delivery ratios as high as that of some of the flooding based protocols. However, they are better steward of network resources and less prone to network congestion.

In this paper, we propose TBR, a new quota based routing protocol for opportunistic networks. TBR introduces a new buffer management strategy and rank the messages in the buffer to schedule the next message to forward or delete. TBR ranks the messages based on message expiry time or TTL, message hop count, message replication count and message size. The use of TTL in ranking the messages allows the message with the earliest deadline to get the preference, while the use of message size allows the shortest message to get the preference. Hop count and replication count are used to ensure network fairness. To evaluate the performance of our protocol, we perform simulation under two different movement models (Keranen et al., 2009)—“Vehicular movement model” and “Random way point movement model” using Opportunistic Network Environment (ONE). We compare our delivery ratios, overheads, and latencies to that of the other popular replication based OpNet routing protocols. Simulation results show that our protocol achieves better delivery ratio than that of all the existing quota based routing protocols while maintaining a low overhead. TBR also achieves better delivery ratios compared to that of the best flooding based protocol with 70 to 75% less overhead and 10–20% less latency.

The rest of the paper is organized as follows. In Section 2, we give a brief background of the existing routing protocols for Opportunistic Networks. Section 3 briefly describes our TBR protocol. Section 4 presents the simulation results and their analysis. Finally, we conclude Section 5 with some future research directions.