performance evaluation of replication-based delay...

Performance Evaluation of Replication-Based Delay-Tolerant Routing Protocols for Underwater Networks

Md. Mahbubur Rahman1 and Muhammad Sajjadur Rahim2 1Department of Computer Science and Engineering

University of Development Alternative (UODA) Dhaka, Bangladesh

2Department of Information and Communication Engineering University of Rajshahi

Rajshahi-6205, Bangladesh Email: [email protected], [email protected]

Abstract—Underwater acoustic channel has extensive and varying propagation delay, high bit error rate and inadequacy of bandwidth. One of the major applications of underwater networks is environmental monitoring and surveillance. To accomplish this task, underwater mobile networks are comprised of Autonomous Underwater Vehicles (AUVs) which roam around the area of surveillance while collecting required data and exchanging among themselves. During their operation the AUVs often suffer from intermittent connectivity as they move around the area and go out of the communication range of other mobile nodes. This feature together with the inherent properties of underwater acoustic channel makes the scenario of underwater mobile networks deployed for monitoring applications appropriate to be modeled as a Delay-Tolerant Network (DTN). Therefore, it is of utmost importance to choose a DTN routing protocol best-suited for underwater mobile networks. In this paper, we have investigated two types of replication-based DTN routing protocols: flooding and Resource Allocation Protocol for Intentional DTN (RAPID) for underwater mobile networks, and their performance is analyzed in terms to packet delivery ratio and average end-to-end delay in different load conditions across various node mobility scenarios. Index Terms—Autonomous underwater vehicles (AUVs), underwater mobile networks, delay-tolerant networking (DTN), flooding, Resource Allocation Protocol for Intentional DTN (RAPID), mobility, routing, simulation, ns2-Miracle

I. INTRODUCTION

Nowadays research in Underwater Acoustic Networks (UANs) has received much attention from both academia and industry because of their important underwater applications for military and commercial purposes. Underwater networks are distinct from terrestrial networks because of the characteristics (large delay, long distance of communication, high bit error rate, and limited bandwidth) of underwater networks as they use acoustic wave instead of electromagnetic wave [1]. Underwater Acoustic Sensor Networks (UASNs) and Autonomous Underwater Vehicle Networks (AUVNs) are two important kinds of UANs. Underwater Acoustic Sensor Networks (UASNs) are composed of many sensor nodes, mostly for a monitoring purpose. Autonomous Underwater

Vehicle Networks (AUVNs) consist of Autonomous Underwater Vehicles (AUVs) that move around an area of operation and interact with one another to collect data and perform collaborative tasks. One of the principle applications of AUVNs is environmental monitoring and surveillance. Here we have considered the characteristics of AUVNs as similar to that of Delay-tolerant-networks (DTNs) [2], where disconnections occur frequently due to propagation phenomena, node mobility, power outages and absence of end-to-end path from source to destination. In DTN environment, common ad hoc routing protocols such as Ad-hoc On Demand Distance Vector (AODV) [3] and Dynamic Source Routing (DSR) [4] fail to establish routes, as they require a route to be established before data packets are actually exchanged; but in the case of DTN, it is not usually possible to discover and set up a route from source to destination before data packets are actually delivered. Most of the nodes in DTN are mobile; hence, the connectivity is established when they come into the transmission range of each other. This leads to a problem how to route a packet from one node to another in DTN. When direct end-to-end paths do not exist, the routing protocols must adhere to the “store-carry-and-forward” approach, which exploits the mobility of nodes to route data. In this approach, data is moved from the source to the next available node and stored there, waiting for an opportunity to forward the data. If present, mobile nodes can carry the stored data while moving, and look for opportunities to forward the data to other nodes towards the destination. These techniques provide eventual data delivery with a certain probability [5] – [7].

On the basis of the number of message copies or replica, routing in DTN is categorized into: forwarding-based and replication-based. Routing protocols that do not replicate a message are called forwarding-based. In forwarding-based routing algorithms, at least one copy of message is maintained in the network. Routing protocols that replicate messages are called replication-based. In replication-based routing algorithms, the source nodes replicate messages to intermediate nodes so that there exists a certain probability of successful message delivery to the destination nodes. Multiple copies of

International Conference on Materials, Electronics & Information Engineering, ICMEIE-2015 05-06 June, 2015, Faculty of Engineering, University of Rajshahi, Bangladesh www.ru.ac.bd/icmeie2015/proceedings/

ISBN 978-984-33-8940--4

mailto:[email protected]

mailto:[email protected]

the same message are created and delivered to a set of nodes called ‘relay nodes’ which store the message until they can contact with the destination node. Forwarding-based routing approaches are generally much less wasteful of network resources, as only a single copy of a message exists in the network at any given time [8], [9], [17]. In this paper, we will present an evaluation of replication-based DTN routing protocols as applied to underwater mobile networks, with focus on the Resource Allocation Protocol for Intentional DTN (RAPID) [10]. In particular, RAPID is a replication-based protocol but replicates packets in controlled fashion and the main purpose of RAPID is to optimize a specific routing metric such as average delay, missed deadlines, or maximum delay. In our simulations, we have compared the standard flooding protocol which plainly forwards a copy of data packet to every node with two instances of RAPID, which focus on the optimization of average delay and maximum delay, respectively.

II. RELATED WORKS

Delay Tolerant Networks, generally known as DTNs, are wireless networks where disconnections may occur frequently due to propagation phenomena, node mobility, power outages and absence of end-to-end path from source to destination. Therefore, Delay -tolerant networks (DTNs) attempt to route network messages via intermittently connected nodes. Many routing techniques have developed for DTNs such as Spray-and-Wait [11] in which only a restricted number of message copies are replicated among nodes, Probabilistic Routing Protocol using History of Encounters and Transitivity (PROPHET) [12], MaxProp [5] routing for vehicle-based disruption-tolerant networks, Resource Allocation Protocol for Intentional DTN (RAPID) [10] to intentionally optimize a single routing metric, such as average end-to-end delivery delay, missed delivery deadlines or maximum delivery delay etc. All of these DTN routing protocols have been developed for ground based terrestrial DTNs. The goal of this paper is to provide an analysis of the performance of DTN routing protocol – RAPID compared to flooding in underwater acoustic mobile networks.

III. REPLICATION-BASED DTN ROUTING PROTOCOLS

UNDER INVESTIGATION

In this section, we give a brief overview on the routing strategy of replication-based DTN routing protocol, and summarize the design of the RAPID protocol. Though forwarding-based routing is much less wasteful of network resources but message delivery ratio is extremely low in many DTN scenarios. On the other hand, replication-based routing protocols achieve higher message delivery ratios [5], since multiple copies of messages exist in the network, and at least one must reach the destination. Balasubramanian et al. proposed a class of replication-based scheme - the Resource Allocation Protocol for Intentional DTN (RAPID) [10] to intentionally optimize a single routing metric.

The standard flooding protocol that we inspect as a base for underwater mobile network scenario works as follows: all

nodes in the network move continuously within the area of operation. As a result, the source node, which store messages or message replicas in its memory, to discover neighbor contacts in its communication range, continuously broadcast PING messages. When the neighbor mobile nodes receive the PING message, they give response to source node by an ACKP message in unicast mode and remain ready to receive the DATA message transmitted by the source node. Upon receiving the ACKP messages, the source node builds a neighborhood list, and then starts transmitting the message until enlisted nodes stay in its communication range. Searching of neighborhood mobile nodes of RAPID is same as the standard flooding protocol. But in the case of replication policy RAPID replicates only those packets that locally result in the highest increase in utility and allocates bandwidth resource to a set of packets in its buffer, in order to optimize the given routing metric.

RAPID [10], which stands for Resource Allocation Protocol for Intentional DTN, is a flooding-based routing protocol. RAPID allocates resources to packets to optimize a specified routing metric- average delay, maximum delay, or missed deadlines. The main feature of RAPID protocol is based on the utility function. RAPID derives a per-packet utility function from the routing metric. RAPID replicates first those packets that locally result in the highest increase in the utility function. The overall protocol consists of the following steps: (i) Metadata is exchanged between nodes in contact to help estimate the packet utilities; (ii) Direct delivery, as packets destined to neighboring final destinations are immediately transmitted; (iii) Replication, as packets are replicated based on the marginal utility, that is the change in utility over the size of the packet; and (iv) Termination, as the protocol ends when contacts break or all packets have been replicated.

IV. SIMULATION RESULTS

In this paper, we concentrate on the performance evaluation of replication-based DTN routing protocols: flooding and RAPID routing protocols in an underwater acoustic mobile network scenario. These two routing protocols are simulated using ns2-Miracle [13]. We have given emphasis on two different instances of RAPID, each pursuing the optimization of a particular objective: the first focuses on the minimization of the average delay (abbreviated as RAPID-AD), whereas the second minimizes the maximum delivery delay (abbreviated as RAPID-MD). The network under consideration is fully mobile, and it is composed of 10 Autonomous Underwater Vehicles (AUVs), where 5 source and 5 destination nodes. All AUVs travel at a given average speed within a shallow-water area of 5 km × 5 km. The transmissions are executed at a bit rate of 4800 bps over a bandwidth of 5 kHz at a center frequency of 11.5 kHz, using the Binary Phase-Shift Keying (BPSK) modulation. Upon the initial exchange of PING and ACKP packets, the source node estimates the distance between itself and the destination node from the round-trip time, and sets its transmit power to match a target packet transmission success rate of 90%. The power budget is calculated according to the empirical equations for path loss and noise [14], [15]. Due to


ISBN 978-984-33-8940--4

computational complexity, usage of a more accurate propagation modeling software suite (as in [18], [19]) has not been considered in this work. In any event, the transmit power never exceeds 190 dB re μPa. The movement is randomized via a Gauss-Markov mobility model [16]. According to this model, each node randomly chooses a movement speed and direction and keeps it constant for a given random time. After this time has expired, a new velocity vector is generated. Unlike a completely random waypoint model, the Gauss-Markov model prescribes that every future random choice is correlated to the previous one according to a correlation parameter 0 ≤ α ≤ 1, where α = 0 means a fully random waypoint selection, and α = 1 is a fully correlated (i.e., linear) movement. If nodes reach the boundaries of the network area, they are automatically bounced off the border. The source nodes generate messages at a rate of 0.25 to 3 packets per minute per node. The overall duration of the simulation is 6 hours, even though one more hour is allowed after the completion of every run, in order to complete the delivery of messages which are still being propagated. All results are averaged over 30 different simulation runs.

Now we start our assessment from Figures 1 and 2, where we respectively report the average packet delivery ratio in the network (defined as the ratio of the number of packets correctly delivered end-to-end to the number of generated packets), and the end-to-end delivery delay (i.e., the time elapsed from when a packet is generated by its source to when it is received by its intended destination). The average movement speed of the nodes is set to 2m/s here.

Figure 1 illustrates the comparison of delivery ratio between RAPID routing protocol and flooding. From the figure we identify that both instances of RAPID perform better than flooding and achieve a delivery ratio of roughly 0.8 even at high packet generation rates. This decent outcome is due to RAPID’s more attentive replica forwarding scheme, which elects whether or not to forward a replica to a given node based on the contribution of that node to the metric to be optimized. Although avoiding to transmit replicas may seem unfavorable in general, it really permits to save time, which can then be dedicated to replicating those messages which exhibit a better chance to fulfill the delay minimization objectives. Figure 2 depicts the comparison of average delivery delay between RAPID routing protocol and flooding. From the figure we observe that pure flooding produces the highest delay than both instances of RAPID, because flooding does not impose any specific law as to how a replica is forwarded, nor does it allocate any priority to the replication of the usually many packets in the queue. In order to examine the effect of node mobility, we varied the average node speed from 2m/s to 8m/s. While we acknowledge that a speed above 4m/s is unlikely for AUVs, we report the related results nevertheless, as this allows to understand whether faster AUVs are beneficial or not in terms of network performance. Figures 3 and 4 respectively present the delivery ratio and delay performance for an average node speed of 6m/s. The general message is that the performance of the network, in fact, improves. In particular, flooding shows a marginally higher delivery ratio and a 10%

lower delivery delay overall: in fact, flooding does not control how suitably the packets are replicated; hence its performance improves with higher node speed, and does not suffer any particular disadvantage because of the shorter encounter duration. RAPID also profits from quicker movement, particularly at low traffic generation rate, thanks to the greater chance to meet the destination of generated packets.

Fig. 1. Delivery ratio as a function of the packet generation rate per node at an average node speed of 2m/s

Fig. 2. Average delivery delay as a function of the packet generation rate per

node at an average node speed of 2m/s

Fig. 3. Delivery ratio as a function of the packet generation rate per node at an

average node speed of 6m/s


ISBN 978-984-33-8940--4

Fig. 4. Average delivery delay as a function of the packet generation rate per

node at an average node speed of 6m/s 1.

To realize a well understanding of the effect of node mobility on the performance metrics of the routing protocols, we report the delivery ratio and delivery delay as a function of the average node speed for different load conditions (i.e., low and high packet generation rates). From Figure 5, we can realize that at low load flooding performs slight healthier than both instances of RAPID in terms of delivery ratio at an average node speed of 2m/s, but as node mobility rises RAPID performs significantly better than flooding. This is because faster mobility makes node meetings more frequent, allowing RAPID a better estimation of inter-meeting times between different pairs of nodes. In turn, the better estimates allow the fine-tuning of the packet replication strategy, whose ultimate effect is a higher packet delivery ratio. Figure 6 shows the performance of average delivery delay at low load between RAPID and flooding. Like previous in this case average delivery delay of RAPID is less than flooding. But both flooding and RAPID enjoy benefits due to increased node mobility in terms of average end-to-end delivery delay, as frequent node meetings contribute to quicker packet delivery to destination, as can be seen from Figure 6.

Fig. 5. Delivery ratio as a function of the average node speed at a packet generation rate of 0.25 pkts/min/node (low load)

Fig. 6. Average delivery delay as a function of the average node speed at a

packet generation rate of 0.25 pkts/min/node (low load) 2.

Figures 7 and 8 illustrate the performance of delivery ratio and average delivery delay at high load between RAPID and flooding routing protocol. From figures we may realize that increase in node mobility in the network area increases the delivery ratio and decrease the average end-to-end delay between source and destination. But if node mobility is so much increase, then decrease the performance because the contact duration is not sufficient for exchanging all required packets.

Fig. 7. Delivery ratio as a function of the average node speed at a packet generation rate of 3 pkts/min/node (high load)

Fig. 8. Average delivery delay as a function of the average node speed at a

packet generation rate of 3 pkts/min/node (high load)


ISBN 978-984-33-8940--4

V. CONCLUSIONS AND FUTURE WORKS

In this research work, we have analyzed the performance of two classes of replication-based DTN routing protocols (flooding and RAPID) in underwater mobile networks. Their performance is evaluated in terms of packet delivery ratio, and average end-to-end delivery delay, in different load conditions and node mobility scenarios. The analysis undoubtedly shows that the RAPID routing protocol gives better results than flooding in both the performance metrics (packet delivery ratio and average end-to-end delivery delay). RAPID offers much improved outcomes for different reasons. Flooding protocol is a simple and does not use any operational policy in its routing strategy and does not control how appropriately the packets are replicated. On the contrary, RAPID is more complex, which uses effective policy in its structure and allows to decrease the number of useless replica transmissions by focusing on those that have a better chance to minimize the delay as well as to maximize the probability of successful delivery.

In future we would like to further explore the performance of other replication-based DTN routing protocols for underwater mobile networks and study their performance in terms of average delivery delay and delivery ratio. We also intend to investigate the performance of forwarding-based DTN routing protocols for underwater mobile networks and to compare them with the replication-based ones. After having insights of both the replication-based and forwarding-based DTN routing protocols, we would focus on designing a DTN routing protocol that would be most suited for underwater mobile networks.

REFERENCES

[1] A. Quazi and W. Konrad, “Underwater acoustic communications”, IEEE Comm. Mag., pp. 24-30, Mar. 1982.

[2] K. Fall, “A delay-tolerant network architecture for challenged internets”, in Proc. of ACM SIGCOMM, Karlsruhe, Germany, Aug. 2003, pp. 27–34.

[3] S. R. Das, C. E. Perkins, E. M. Royer, and M. K. Marina, “Performance comparison of two on-demand routing protocols for ad hoc networks”, IEEE Personal Comm. Mag., vol. 8, no. 1, pp. 16–28, Feb. 2001.

[4] D. B. Johnson and D. A. Maltz, Mobile Computing. Kluwer Academic Publishers, Feb. 1996, ch. Dynamic source routing in ad hoc wireless networks, pp. 153–181.

[5] J. Burgess, B. Gallagher, D. Jensen, and B. N. Levine, “MaxProp: routing for vehicle-based disruption-tolerant networks”, in Proc. of IEEE INFOCOM, Barcelona, Spain, Apr. 2006.

[6] P. Juang, H. Oki, Y. Wang, M. Martonosi, L. S. Peh, and D. Rubenstein, “Energy-efficient computing for wildlife tracking: design tradeoffs and early experiences with zebranet”, in Proc. of ASPLOS, San Jose, CA, 2002, pp. 96–107.

[7] A. Chaintreau, P. Hui, J. Crowcroft, C. Diot, R. Gass, and J. Scott, “Impact of human mobility on opportunistic forwarding algorithms”, IEEE Trans. on Mob. Comp., vol. 6, no. 6, pp. 606–620, Jun. 2007.

[8] S. Jain, K. Fall, and R. Patra, “Routing in a delay-tolerant network”, in Proc. of ACM SIGCOMM, Portland, OR, 2004, pp. 145–157.

[9] D. Henriksson, T. F. Abdelzaher, and R. K. Ganti, “A caching-based approach to routing in delay-tolerant networks”, in Proc. of ICCCN, Honolulu, HI, Aug. 2007, pp. 69–74.

[10] A. Balasubramanian, B. N. Levine, and A. Venkataramani, “DTN routing as a resource allocation problem”, in Proc. of ACM SIGCOMM, Kyoto, Japan, Aug. 2007, pp. 373–384.

[11] T. Spyropoulos, K. Psounis, and C. S. Raghavendra, “Spray and Wait: an efficient routing scheme for intermittently connected mobile networks”, in Proc. of ACM WDTN, Philadelphia, PA, Aug. 2005, pp. 252–259.

[12] A. Lindgren, A. Doria, and O. Scheln, “Probabilistic routing in intermittently connected networks”, ACM Mobile Comp. and Comm. Review, vol. 7, no. 3, pp. 19–20, Jul. 2003.

[13] N. Baldo, F. Maguolo, M. Miozzo, M. Rossi, and M. Zorzi, “NS2-MIRACLE: a modular framework for multi-technology and cross-layer support in network simulator 2”, in Proc. of ACM NSTools, Nantes, France, Oct. 2007.

[14] R. Urick, Principles of Underwater Sound, New York: McGraw-Hill, 1983.

[15] M. Stojanovic, “On the relationship between capacity and distance in an underwater acoustic communication channel”, ACM Mobile Comp. and Comm. Review, vol 11, pp. 34-43, Oct. 2007.

[16] B. Liang and Z. J. Haas, “Predictive distance-based mobility management for PCS networks,” in Proc. of IEEE INFOCOM, New York, NY, Mar. 1999, pp. 1377–1384.

[17] Z. Guo, B. Wang, and J. Cui, “Prediction assisted single-copy routing in underwater delay tolerant networks,” in Proc. of IEEE GLOBECOM, Miami, FL, Dec. 2010.

[18] F. Guerra, P. Casari, and M. Zorzi, “World Ocean Simulation System (WOSS): a simulation tool for underwater networks with realistic propagation modeling,” in Proc. of ACM WUWNet 2009, Berkeley, CA, Nov. 2009.

[19] R. Masiero, S. Azad, F. Favaro, M. Petrani, G. Toso, F. Guerra, P. Casari, and M. Zorzi, “DESERT underwater: an ns-miracle-based framework to design, emulate and realize test-beds for underwater network protocols,” in Proc. of IEEE Oceans, Yeosu, Korea, May 2012.


ISBN 978-984-33-8940--4

performance evaluation of replication-based delay...

Documents