Particle swarm optimization in intelligent routing of delay-tolerant network routing
© Omidvar and Mohammadi; licensee Springer. 2014
Received: 4 July 2014
Accepted: 29 August 2014
Published: 11 September 2014
Delay-tolerant networks (DTNs) are wireless partitioned networks. Because of intermittency, mobile ad hoc network (MANET) routing protocols are not efficient in DTNs. Wildlife tracking, vehicular networks, interplanetary networks, etc. are different applications of DTN. Regarding DTN applications, different parameters should be considered while designing DTN routing protocols. Message delivery ratio, message delivery delay, overhead, message drop, etc. are some important factors that are usually considered in routing algorithms. This paper proposes a method which tries to reduce overhead and message drop while increasing message delivery ratio. Choosing the appropriate number of message copies to distribute in the network is important. Few numbers of copies can lead to message drop. So, the message cannot be delivered to the destination. On the other hand, increasing the number of copies causes overhead increase in the network. The proposed algorithm uses particle swarm optimization (PSO) in intelligent choosing of number of message copies. Regarding message delivery ratio and network overhead, PSO greatly helps in finding the suitable number of copies. In order to evaluate our method, which is called PSODTN, we compared it with epidemic routing (ER) and probabilistic routing protocol using history of encounters and transitivity (PROPHET). PSODTN helps to reduce overhead, on average, 95.6% compared to ER and PROPHET. While reducing overhead, PSODTN message delivery ratio is on average 98%.
Delay-tolerant networks (DTNs) are wireless ad hoc networks with intermittent connections, limited power and radio range, long variable delays, etc. These networks have different applications such as interplanetary networks[2, 3], military networks, vehicular networks, wildlife tracking, etc.
Traditional mobile ad hoc network (MANET) routing protocols are not efficient in DTNs because of intermittency. DTN uses asynchronous connections for delivering messages. Store-carry-forward (SCF) is the approach used in delivering data. The node stores the packet in its buffer and carries it around until finding the next appropriate hop for forwarding messages[7–9].
Different approaches have been suggested, but there is still a need to find routing approaches which have better performance. Finding the appropriate number of copies is important in DTNs. The number of message copies should be selected such that, while reducing overhead, it keeps message delivery ratio in a good level. This paper tries to find the number of message copies using particle swarm optimization (PSO). The proposed method, which is called PSODTN, considers message delivery ratio and overhead to find the number of message copies. The rest of the paper is organized as follows. Section 2 discusses related works. The proposed method is stated in Section 3. PSODTN simulations are presented in Section 4. Section 5 concludes the paper.
2 Related works
Routing in DTN has great importance because of intermittency. Different routing approaches have been proposed[10, 11] which are classified from different aspects. Based on the information DTN routing protocols use from the network, they can be categorized as forwarding based, history based, and flooding.
Forwarding-based approaches use a single copy of the message to forward. This helps to reduce overhead. On the other hand, message delivery delay will increase greatly. Minimum estimated expected delay (MEED) is an example of this group. Six types of forwarding routing are presented in.
In history-based approaches, routing decisions are made based on the history of encounters between nodes. Probabilistic routing protocol using history of encounters and transitivity (PROPHET) is one of the well-known algorithms of this category. PROPHET uses delivery predictability, a parameter showing the predictability of encountering a node to the destination based on its past encounters with other nodes. MaxProp is another approach which considers encounters with other nodes but not necessarily the destination.
The FResher Encounter SearcH (FRESH) is another prediction-based approach which considers time passed since the last encounter of every node with other nodes. Resource allocation protocol for intentional DTN (RAPID) is also another example of this approach.
Another effective usage of history-based approaches is social relations. These approaches consider social similarities. These methods consider social contacts in addition to predicting future movements. Label, SimBet, Bubble Rap, and SocialCast are instances of these social predictive methods. Label algorithm was the first approach to use social characteristics in opportunistic networks. This algorithm uses labels to improve message delivery ratio. SimBet is used in networks which are clustered and in which nodes cannot meet the destination. Bubble Rap forwards messages using node centrality and community structure. SocialCast considers destination interests in addition to moving patterns and social connections.
Flooding algorithms, such as epidemic routing, are the most reliable methods in delivering messages. Each node, which has the message, gives a copy of the message to every node it encounters. These methods waste network resources such as energy, bandwidth, etc. They cause overhead in network and overflow in buffer.
Recently, there have been many researches done on DTN routing. Vasilakos et al. have studied delay-tolerant networks and their various protocols and applications in. Spyropoulos et al. have tried to investigate network characteristics that are relevant to routing processes. Regarding different applications of DTN, Sun et al. considered performance of DTN protocols in space communication. Routing in vehicular delay-tolerant network (VDTN) as one of the DTN applications has also been studied by researchers. Zeng et al. considered energy optimization in VDTN routing. In addition to DTN routing challenges, some features in wireless networks make routing conditions more difficult. Youssef et al. have discussed routing metrics in cognitive radio networks. They have prepared challenging metrics in wireless networks such as spectrum availability, interruption time, etc. Channel assignment in wireless networks is another problem discussed in. Cheng et al. have used PSO for channel assignment in wireless mesh networks. Considering the researches done, there is still a need to optimize network resource usage.
Regarding limited network resources, the proposed method tries to reduce network overhead while keeping message delivery ratio in a good range. The next section discusses the proposed method.
3 Proposed method
Node resources such as buffer, energy, etc. are limited. Due to these problems, routing in DTNs is challenging. When designing DTN protocols, different parameters should be considered. Various applications need different parameters to be tuned. In some cases, designers want to improve message delivery ratio while reducing overhead. Some others aim to reduce message delivery delay, etc. Increasing the number of message copies helps to improve message delivery ratio while increasing overhead and message drop. On the other hand, reducing the number of message copies helps to decrease overhead while increasing message delivery delay and reducing message delivery ratio. So, the number of message copies distributed in the network has high importance, which will be discussed in this paper.
In this paper, by means of controlling the number of message copies, we try to reduce overhead while maintaining message delivery ratio in a good range (delivery ratio >90%). Since in DTN, increasing the number of message copies delivered to the destination is important, increasing message delivery ratio as much as possible is essential for us. We use PSO to choose the appropriate number of message copies.
The proposed algorithm uses PSO. PSO is a successful method among evolutionary algorithms (EA) to solve optimizing problems. It was first developed by Kennedy and Eberhart in 1995. At first, it was designed to simulate social systems such as bird flocks. In PSO, each possible solution, called particle, is assigned location and velocity. Each particle stores its location and fitness. Pbest is assigned to the best solution locally found. Gbest is given to the particle globally optimized. Using PSODTN, the number of message copies to distribute in the network is found.
Message delivery ratio is defined as the ratio of delivered messages to the created ones.
In this approach, we use message delivery ratio as the fitness function. PSODTN tries to reduce overhead while maintaining message delivery ratio in a good stage (message delivery ratio >90%).
As was mentioned before, SCF is used in DTNs because intermittency in DTNs makes conventional MANET routing protocols inefficient. Choosing the appropriate number of message copies has great importance since large numbers cause great overhead in the network and few numbers of copies can lead to message drop. When there are few message copies in the network and message time to live (TTL) gets 0, the message will be dropped. PSODTN tries to select enough copies considering message delivery ratio while reducing message overhead.
In (2), because of random waypoint (RWP) mobility used in our simulations, ω = 1.3683.
Tpause, which states the average pause after an epoch, belongs to [0,Tmax].
In calculating the delay of PSODTN, we have to consider PSO delay in addition to the delay caused by sparse networks.
DPSO shows the delay caused by PSO in finding the appropriate number of copies.
4 PSODTN simulation
Opportunistic network simulator environment (ONE) is used to evaluate the proposed method. ONE is a Java-based software developed in the University of Helsinki in Finland. Simulations have been repeated for 50 times and results are averaged.
In our simulations, c 1 = c 2 = 1. We have studied different swarm size population to choose the appropriate size.
Two scenarios were implemented to evaluate the proposed method.
Parameters used in simulating scenario 1
500 × 500
As the population increases, message delivery ratio increases while increasing overhead. Average hop count will also increase. Regarding this study, we set the population size to 30.
Parameters used in simulating scenario 2
500 × 500
Considering the results of the first scenario, these simulations prove PSODTN success in comparison to ER and PROPHET.
Considering both scenarios, overhead has reduced by an average of 95.6% compared to ER and PROPHET while we have a message delivery of 98%.
This trial also proves PSODTN success in reducing overhead while maintaining message delivery ratio in a good stage.
Delay-tolerant networks are sparse networks with intermittent connections. DTNs have different applications such as wildlife tracking, military battle field networks, etc. Conventional MANET routing protocols are not useful in these networks because of intermittency. SCF is used in these networks for transferring messages. Finding the appropriate number of copies distributed in the network is very important since fewer ones lead to message drop and large numbers cause huge overhead. In this paper, PSO is used to find the appropriate number of message copies. This helps to reduce network overhead while maintaining message delivery ratio in a good range. Two scenarios were tested to evaluate PSODTN. Considering both scenarios, overhead has reduced on average by 95.6% compared to ER and PROPHET. Message delivery ratio in PSODTN is on average 98%. Simulations prove PSODTN success in finding the appropriate number of message copies.
Future works should consider other optimization methods. Comparing different optimization methods in finding the appropriate number of copies helps to choose the method which has less overhead accompanied with more message delivery ratio. In addition to optimizing overhead, buffer usage and message delivery delay should also be considered for optimization in future works. Designing hybrid methods to solve these multi-objective optimization approaches should be evaluated in future works.
- Fall K: A delay-tolerant network architecture for challenged Internets. In Proc. ACM 2003 Conf. on Applications, Technologies, Architectures, and Protocols for Computer Communications, ser. SIGCOMM’03. ACM, New York, USA; 2003:27-34.Google Scholar
- Rango FD, Tropea M, Laratta GB, Marano S: Hop-by-hop local flow control over interplanetary networks based on DTN architecture. In Proc. IEEE IC 2008. IEEE, Beijing, China; 2008:1920-1924.Google Scholar
- Lebrun J, Chuah C-N, Ghosal D, Zhang M: Knowledge-based opportunistic forwarding in vehicular wireless ad hoc networks. In Proc. IEEE Vehicular Tech. Conference, Vol. 4. IEEE, Stockholm, Sweden; 2005:2289-2293.Google Scholar
- Disruption Tolerant Networks Program. . Accessed Jan 2009 http://www.darpa.mill/ato/solicit/DTN/
- Burleigh S, Hooke A, Torgerson L, Fall K, Cerf V, Durst B, Scott K: Delay-tolerant networking: an approach to interplanetary internet. IEEE Comm. Mag. 2003, 41: 128-136.View ArticleGoogle Scholar
- Juang P, Oki H, Wang Y, Martonosi M, Peh LS, Rubenstein D: Energy-efficient computing for wildlife tracking: design tradeoffs and early experiences with Zebranet. In Proc. SIGPLAN. ACM, San Jose, USA; 2002:96-107.Google Scholar
- Jain S, Fall K, Patra R: Routing in a delay-tolerant network. In Proc. Conf. Appl., Technol., Architectures, Protocols Comput. Commun. ACM, Portland, USA; 2004:145-158.Google Scholar
- Zhang Z: Routing in intermittently connected mobile ad hoc networks and delay-tolerant networks: overview and challenges. Commun. Surv. Tuts. 2006, 8(1):24-37.View ArticleGoogle Scholar
- Fall K, Farell S: DTN: An architectural retrospective. IEEE J. Sel. Areas Commun. 2008, 26(5):828-836.View ArticleGoogle Scholar
- Spyropoulos T, Psounis K, Raghavendra CS: Spray and wait: an efficient routing scheme for intermittently connected mobile networks. In Proc. SIGCOMM WDTN. ACM, Philadelphia, USA; 2005:252-259.View ArticleGoogle Scholar
- Wang Y, Jain S, Martonosi M, Fall K: Erasure-coding based routing for opportunistic networks. In Proc. SIGCOMM’05 Workshops, August 22–26, Philadelphia, USA. ACM; 2005:229-236.Google Scholar
- Jones EPC, Li L, Ward PAS: Practical routing in delay-tolerant networks. In Proc. ACM SIGCOMM WDTN. ACM, Philadelphia, USA; 2005:237-243.Google Scholar
- Spyropoulos T, Psounis K, Raghavendra CS: Efficient routing in intermittently connected mobile networks: the single-copy case. IEEE/ACM Trans. Netw. 2008, 16(1):63-76.View ArticleGoogle Scholar
- Lindgren A, Doria A, Scheln O: Probabilistic routing in intermittently connected networks. Lect Notes Comput Sci 2004, 3126: 239-254. 10.1007/978-3-540-27767-5_24View ArticleGoogle Scholar
- Burgess J, Gallagher B, Jensen D, Levine BN: MaxProp: routing for vehicle-based disruption-tolerant networks. In Proc. INFOCOM. (IEEE, Barcelona, Spain; 2006:1-11.Google Scholar
- Dubois-Ferriere H, Grossglauser M, Vetterli M: Age matters: efficient route discovery in mobile ad hoc networks using encounter ages. In Proc. ACM MobiHoc. ACM, Annapolis, USA; 2003:257-266.Google Scholar
- Balasubramanian A, Levine BN, Venkataramani A: Replication routing in DTNs: a resource allocation approach. IEEE/ACM Trans. Netw. 2010, 18(2):596-609.View ArticleGoogle Scholar
- Hui P, Crowcroft J: How small labels create big improvements. In Proc. IEEE PERCOM Workshops. IEEE, White Plains, USA; 2007:65-70.Google Scholar
- Daly EM, Haahr M: Social network analysis for routing in disconnected delay-tolerant MANETs. In Proc. ACM MobiHoc. ACM, Montreal, Canada; 2007:32-40.Google Scholar
- Hui P, Crowcroft J, Yoneki E: Bubble rap: social-based forwarding in delay tolerant networks. In Proc. ACM MobiHoc. ACM, Hong Kong, China; 2008:241-250.Google Scholar
- Costa P, Mascolo C, Musolesi M, Picco GP: Socially-aware routing for publish-subscribe in delay-tolerant mobile ad hoc networks. IEEE J. Sel. Areas Commun. 2008, 26(5):748-760.View ArticleGoogle Scholar
- Vahdat A, Becker D: Epidemic routing for partially connected ad hoc networks, Duke University. Tech. Rep 2000. CS.-2000-06Google Scholar
- Vasilakos A, Zhang Y, Spyropoulos TV: Delay Tolerant Networks: Protocols and Applications. CRC Press, Boca Raton; 2012.Google Scholar
- Spyropoulos T, Naveed Bin Rais R, Turletti T, Obraczka K, Vasilakos A: Routing for disruption tolerant networks: taxonomy and design. Wireless Netw 2010, 16(8):2349-2370. 10.1007/s11276-010-0276-9View ArticleGoogle Scholar
- Sun X, Yu Q, Wang R, Zhang Q, Wei Z, Hu J, Vasilakos A: Performance of DTN protocols in space communications. Wireless Netw. 2010, 19(8):2029-2047.View ArticleGoogle Scholar
- Zeng Y, Xiang K, Li D, Vasilakos A: Directional routing and scheduling for green vehicular delay tolerant networks. Wireles Netw. 2013, 19: 161-173. 10.1007/s11276-012-0457-9View ArticleGoogle Scholar
- Youssef M, Ibrahim M, Abelatif M, Chin L, Vasilakos A: Routing metrics of cognitive radio networks: a survey. IEEE Commun. Surv. Tuts. 2014, 16(1):92-109.View ArticleGoogle Scholar
- Cheng H, Xiong N, Vasilakos A, Yang LT, Chen G, Zhuang X: Nodes organization for channel assignment with topology preservation in multi-radio wireless mesh networks. Ad Hoc Networks 2012, 10(5):760-773. 10.1016/j.adhoc.2011.02.004View ArticleGoogle Scholar
- Kennedy J, Eberhart RC: Particle swarm optimization. In Proc. IEEE International Conference on Neural Networks, Vol. 4. Perth, Australia; 1995:1942-1948.View ArticleGoogle Scholar
- Groenevelt R, Nain P, Koole G: The message delay in mobile ad hoc networks. Elsevier J. Perform. Eval. 2005, 62: 210-228. 10.1016/j.peva.2005.07.018View ArticleGoogle Scholar
- Groenevelt R: Stochastic Models in Mobile Ad Hoc Networks. University of Nice Sophia, PhD dissertation; 2005.Google Scholar
- Spyropoulos T, Psounis K, Raghavendra CS: Performance analysis of mobility-assisted routing. In Proc. ACM MobiHoc. ACM, Los Angeles, USA; 2006:49-60.Google Scholar
- Yoon J, Liu M, Noble B: Random waypoint considered harmful. In Proc. IEEE INFOCOM. IEEE, San Francisco, USA; 2003:1312-1321.Google Scholar
- Bettstetter C, Hartenstein H, P’erez-Costa X: Stochastic properties of the random waypoint mobility model: epoch length, direction distribution, and cell change rate. In Proc. MSWIM. ACM, New York, USA; 2002:7-14.Google Scholar
- TKK/COMNET: Project page of the ONE simulator. . Accessed Jan 2012 http://www.netlab.tkk.fi/tutkimus/dtn/theone/
- Kerӓnen A, Kӓrkkӓinen T, Ott J: Simulating mobility and DTNs with the ONE. J. Commun. 2010, 5(2):92-105.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.