• No results found

A Dynamic Mesh-based Wireless Multicast Opportunistic Routing with Network Coding

N/A
N/A
Protected

Academic year: 2021

Share "A Dynamic Mesh-based Wireless Multicast Opportunistic Routing with Network Coding"

Copied!
6
0
0

Loading.... (view fulltext now)

Full text

(1)

Energy Procedia 16 (2012) 952 – 957

1876-6102 © 2011 Published by Elsevier B.V. Selection and/or peer-review under responsibility of International Materials Science Society. doi:10.1016/j.egypro.2012.01.152

Energy

Procedia

Energy Procedia 00 (2011) 000–000

www.elsevier.com/locate/procedia

2012 International Conference on Future Energy, Environment, and Materials

A Dynamic Mesh-based Wireless Multicast Opportunistic

Routing with Network Coding

Daru Pan

*, Zhaohua Ruan, Wei Cao

School of Physics and telecommunication, South China Normal University, Guangzhou, China

Abstract

This paper presents a practical multicast routing, which dynamically transforms a conventional multicast tree into a multicast mesh and increases the opportunities for coding by applying double direction network coding scheme. The simulation shows the proposed algorithm is a practical multicast routing, and it has better performances by comparing with ODMRP and MORE-M in terms of coding gain and packet delivery ratio.

© 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of [name organizer] Keywords: Network coding; Opportunistic Routing; Multicast.

1. Main text

Wireless Multicast is a mechanism that delivers information from one source to a set of destinations simultaneously in an efficient manner. It is widely used in many group-oriented applications including disaster management, battlefields, audio/video conferencing, e-commerce, e-education, etc. These applications demand efficient and reliable multicast routes under different wireless network environment, and reliable multicast requires that every receiver receives the correct information sent by the source. Early efforts in multicast routing in a wireless ad-hoc network were to adapt traditional tree-based algorithms such as MAODV[1], AMRoute [2], and AMRIS [3]. However, this class of protocols does not

perform well for the mobility of nodes. Another category of protocols including ODMRP [4] and CAMP [5]

was designed to build routing meshes rather than routing trees. There are several paths between a source and a receiver in a routing mesh. When nodes move, one of the paths between a source and a receiver may break, but duplicate copies of a packet are still able to be delivered to the destination via alternate paths. This would result in better packet delivery ratios than tree-based algorithms. So ODMRP suffers only

* Corresponding author. Daru Pan E-mail address: pandr@scnu.edu.cn.

© 2011 Published by Elsevier B.V. Selection and/or peer-review under responsibility of International Materials Science Society.

Open access under CC BY-NC-ND license.

(2)

Pan Daru*, Ruan Zhaohua, Cao Wei / Energy Procedia 00 (2011) 000–000

minimal data packets loss, and has low channel and storage overhead because of it on-demand nature and multi-paths available. However, its sender-initiated mesh construction method results in larger mesh and numerous unnecessary transmissions of data packets compared to a receiver-initiated mesh construction approach. ODMRP with multi-point relay (ODMRP-MPR)[6] reduces the packet over- head using

multi-point relay nodes. The multi-multi-point relay nodes minimize the broadcast overhead by reducing duplicated packet forwarding, which brings high scalability and effectively solves the unidirectional link problem of wireless communication.

Network coding is a technique that successfully increases transmission capacity of a network by mixing together data from different sources and broadcasting the coded data. Since proposed by Ahlswede et al. [7], Network Coding has seen the application in areas of routing, security as well as error

recovery and failure resilience. For the problem of multicast under wireless network environment, using network coding at the intermediate nodes can achieve the network capacity. Recently, network coding in wired networks with stable topologies has been concerned extensively, but the research and application of network coding in wireless networks are still in the primary stage. COPE[8] is the first practical

implementation of network coding to achieve higher throughput in multi-hop wireless networks unicast routing, while in COPE, the paths of flows are decided by the routing protocol, which means the packets goes through a fixed route, and this might lead to no coding opportunities. To solve this problem, several opportunistic routing algorithms [ExOR [9], MORE [10], BEND [11]] have been proposed. Unlike traditional

routing chooses the next hop before transmitting a packet, ExOR uses an opportunistic routing to allow any node that overhears the transmission and is closer to the destination to forward the packet. However, ExOR also introduces some difficult challenge. It mixes the MAC and the routing to prevent multiple nodes, which may hear a packet broadcast from unnecessarily forwarding the same packet. Szymon Chachulski et al. presented a MAC-independent opportunistic routing protocol in MORE, which randomly mixes packets before forwarding them. The highlight of MORE is that it needs no special scheduler to coordinate routers, and it runs directly on the top of 802.11 MAC protocol. While in MORE, packets from the same source are divided into batches, and only the packets in the same batch can be coded together, this means only the packets form the same source can be coded together, as a result, this will lose some opportunities of coding. BEND combines the features of network coding and opportunistic forwarding in 802.11-based mesh networks to create more coding opportunities in the network. While the coding opportunistic position lies in the nodes or its neighbors on route, which prevent the forwarder from fully exploiting spatial reuse, and the packets in the same flow direction cannot be coded in BEND.

In this paper, we propose a mesh-based wireless multicast opportunistic routing with double direction network coding. Where we keep the packet sender combine the packets randomly in a local area along the route, create more coding opportunities by introducing double direction coding and dynamically transform a conventional multicast tree into a multicast mesh.

2. Design of Mesh-based Wireless Multicast Opportunistic Routing

2.1. Network Model

The topology of the wireless network is modeled as a directed hyper graph G V E=( , ). V is the set of nodes, and E is the set of hyper arcs. Each node in the network can be a source or destination of traffic. A hyper arc is a pair ( , )I J , where I , the start node, is an element of V, and J, the set of end nodes, is a nonempty subset of V. Each hyper arc ( , )I J represents a broadcast link from node I to nodes in J, which indicates that a packet transmitted by node I may be received by nodes in J due to the broadcast nature of the wireless channel. For node I, letIn I( ) denote the set of incoming channels to I and Out I( ) the set

(3)

of outgoing channels. A set of unicast sessions flow S= 

{

s1,...,sn

}

is transmitted through the network.

There are

n

commodities in the network corresponding to each unicast session.

2.2. Dynamic Mesh Forming.

In ODMRP, The multicast mesh nodes are formed by using two types of control packet: Join Query and Join Reply. Whenever a multicast group member desires to send packets to other members, the request phase begins. In the request phase, the source broadcasts a packet called JOIN REQUEST periodically to the entire network that acts as member advertising packet. Different from the ODMRP, in this proposed Mesh-based Wireless Multicast Opportunistic Routing, the multicast mesh is formed automatically, where the broadcast nature of the wireless signaling is used. Before sending packets, the source constructs a multicast tree from the source to all its multicast receivers according to the conventional multicast routing algorithm, which is a shortest-ETX tree. The routing metric is expected transmission count, descried as following:

EXT=1−1d (1)

1 (1 f)(1 r)

d= − −dd (2)

Where d,df and dr refer to over-all packet loss rate, data packet loss rate and acknowledge packet

loss rate.

When a node transmits the packets, there is a chance for the node closer than the next hop to the destination overhears the packets, and there is also a chance for the destination overhears the packets from the nodes which are not so closer to the destination. In this proposed scheme, all nodes in the original multicast tree and their neighbor nodes can act as forwarders, and they are not necessary to be the next hop of the sender for forwarding packets because of the spatial reuse. Once a node overhears a packet from a multicast tree, and finds it has a better EXT than the next hop of the sender, it adds itself to the multicast group. Thus, the original multicast tree transforms into a multicast mesh structure, and this multicast mesh is dynamic.

2.3. Double Direction Network Coding and Decoding

The flow of the scheme is shown in Fig.1.

Firstly, the source node A builds up a shortest-ETX multicast tree according to the conventional multicast routing algorithm. M packets are originated and divided into several batches at a source node A. a packet

p

i from the first batch is in the form of a k-dimensional row vector. This k-dimensional linear network code consists of a scalar

L

,i j for every hyper arc pair( , )i j and the symbols

p f

i

.

d,. So the

source randomly sends a coded packet via the linear formula.

, ( . )

j i j i d i

p =

L p f (3)

Where Lj=lj1..., ,...,lji ljk is the code vector of the packet, which describes how to generate the coded

packet from the native packets and how to decode by the destination node when receive all the

k

corresponding linear coded packets.

Secondly, when a node b hears a packet, it checks whether the received coded packet contains any new information, or technically, whether the packet is linear independent from the packets the node has previously received from this batch and EXTb>EXTnexthop , if so, the node adds the this node to the

(4)

Pan Daru*, Ruan Zhaohua, Cao Wei / Energy Procedia 00 (2011) 000–000

random linear combination to the received coded packet with a packet it has heard from the same batch and broadcasts it.

Packets spread along the giving shortest-ETX multicast tree

If the neighbor nodes hear a packet, EXTb>EXTnexthop,

and there are some coding opportunities.

No Add this nodes to the multicast

group to form a mesh. linear code packets from the same

batch with the code vector and broadcast the coded packets

Yes

If a. The next-hop receiver of p1 is p2’s previous forwarder or one of its

neighbors. b. The next-hop receiver of p2 is p1’s

previous forwarder or one of its neighbors.

combine two packets from different batches in different flow directions

by applying a simple XOR ⊕operation and broadcast it

Yes M packetes are originated and divided into several batches at a source node A. Node A builds up a shortest-ETX

multicast tree

No Directly deliver packets to next Node along the multicast tree

decoded the batch of packets If the receiver is

destination? Yes No

Decode the Inter-flow coded packet

Fig.1. Flow Chart of the Proposed Routing

It can easily proved this linear combination of coded packets is also a linear combination of native packets: , ' ( . ) j i j i d i p =∑L p f (4) Where pi is the native packet, and pjis the heard coded packet

Then j' j j( . )ji d j j i ji( . )i d i (j j j)( . )i d

p =∑r p f =∑ ∑r α p f =∑ ∑rα p f

(5)

Where

r r

j

=

j1

..., ,...,

r

ji

r

jkis the code vector of coded packetpj'. We thus prove that it is a linear

combination of native packets.

After that, the node can further combine two packets from different batches in different flow directions by applying a simple exclusive OR operation, and the decoding of this packet will be done in the next hope by simply apply applies the same exclusive OR operation again.

Finally, when a node receives a packet, it checks if it is the destination. It will keep waiting until it receives all the K linear independent packets, and then decoded the batch of packets by using the matrix inversion of the Code Vector.

3. Simulation and Analysis

We implement the proposed routing by modifying and developing the traditional routing module. We compare it with traditional mesh multicast routing ODMRP, and simulated MORE-M according to the packet delivery ratio and coding gain. The simulated environment consists of 100 wireless nodes placed randomly in a 1200×800 m2 area with a maximum node speed of 20m/s. The radio propagation range is

250m, and the channel capacity is 2Mbps.

Fig 2 shows coding gain of the proposed routing and simulated MORE-M under different multicast group sizes. The coding gain refers to the ratio of the number of transmissions required by the traditional best routing with no-coding approach, to the minimum number of transmissions used by the proposed routing to deliver the same set of packets. By definition, this number is greater than or equal to 1. As the multicast group size increases, the coding gains of both schemes are increased. This is because more group members bring more opportunistic coding for them. When the group size is larger than 10, the proposed routing acquires more coding opportunities than MORE-M by using the double direction coding. This makes it more likely for the proposed algorithm to transmit more packets at a time, and have more chances to combine packets form the same flow and from different flows.

multicast group size

0 5 10 15 20 25 30 35 C odi ng G ai n 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6

The proposed routing Simulated MORE-M

multicast traffic load( Mb/sec)

0 2 4 6 8 10 12 A ver age P ac ket D el iv er y R at io( % ) 0.5 0.6 0.7 0.8 0.9 1.0

The proposed routing ODMRP Simulated MORE-M

Fig. 2. Coding gain of different algorithms. Fig. 3. Packet delivery ratio of different algorithms

Fig 3 compares the packet delivery ratio of the proposed algorithm with ODMRP and MORE-M. The delivery ratio presents the ratio of the number of packets received by multicast receivers versus the number of data packets supposed to be received. The corresponding value shows the effectiveness of the protocol in handling the traffic and delivering the data to the intended multicast receivers. For all kinds of traffic load, all schemes performance is affected by the increasing traffic load. The proposed mesh-based multicast routing and ODMRP have the better packet delivery ratio than MOER-M. Because both are using a mesh multicast structure, which can bring the alternative route for the packets to deliver. In low

(6)

Pan Daru*, Ruan Zhaohua, Cao Wei / Energy Procedia 00 (2011) 000–000

traffic load, the proposed routing is comparable with ODMRP. As the traffic load increases, the packet delivery ratio of ODMRP drops faster than the proposed routing. The dynamic mesh structure enables the proposed routing to adapt to the high-traffic load and suffer only minimal data loss.

4. Conclusions

In this paper, we proposed a practical mesh-based multicast routing with double direction network coding. The proposed algorithm utilizes the broadcasting nature of wireless signaling to create coding opportunities for the wireless network. It increases the packet delivery ratio by building a dynamic multicast mesh structure, where nodes are allowed to add to the multicast group if it has better opportunity to transmit the packets. The proposed algorithm also introduces double direction network coding to improve the ability of transmission. Packets from the same flow and from different flows can code and decode in a different way. The simulation was performed to prove the effectiveness of the proposed method under various traffic load and multicast group sizes. The proposed routing outperforms other routing protocols with respect to the packet delivery ratio and the coding gain.

Acknowledgements

This work is supported by the National Natural Science Foundation of China under grant No.61172087, and the Natural Science Foundation of Guangdong Province under grant No. 06300923

References

[1] E.M. Royer, C.E. Perkins, Multicast operation of the ad hoc ondemand distance vector routing protocol, in: ACM MobiCom, 1999.

[2] Jason Xie, Rajesh R. Talpade, Anthony McAuley, Mingyan Liu, AMRoute: ad hoc multicast routing protocol, ACM/Kluwer Mobile Networks and Applications 7 (6) (2002).

[3] C.W. Wu, Y.C. Tay, C.K. Toh, Ad hoc multicast routing protocol utilizing increasing id-numberS (AMRIS) functional specification, in:IEEE MILCOM, 1999, pp. 25–29.

[4] S.J. Lee, W. Su, M. Gerla, On-demand multicast routing protocol in multihop wireless mobile networks, ACM/Kluwer Mobile Networks and Applications 7 (6) (2002).

[5] J.J. Garcia-Luna-Aceves, E.L. Madruga, The core-assisted mesh protocol, IEEE Journal on Selected Areas in Communications 17 (8) (1999) 1380.

[6] Ruiz PM, Gomez-Skarmeta AF. Reducing data-overhead of mesh-based ad-hoc multicast routing protocols by Steiner tree meshes. In: Proceedings of the first IEEE communications society conference on sensor and ad hoc communica- tions and networks (SECON); 2004.

[7] R. Ahlswede, N. Cai, S.R. Li, and R. W. Yeung, Network information flow, IEEE Trans on Information Theory, 2000, 46(4):1204-1216.

[8] S. Katti, H. Rahul, W. Hu, D. Katabi, M. Medard, and J. Crowcroft, “XORs in The Air: Practical Wireless Network Coding,” in Proceedings of ACM SIGCOMM 2006.

[9] S. Biswas and R. Morris, Opportunistic routing in multi-hop wireless networks. In SIGCOMM, 2005.

[10] S. Chachulski, M. Jennings, S. Katti and D. Katabi, Trading Structure for Randomness in Wireless Opportunistic Routing , in SIGCOMM’07.

[11] J. Zhang, Y. P. Chen and I. Marsic, Network Coding Via Opportunistic Forwardingin Wireless Mesh Network, In Proceedings of WCNC2008.

References

Related documents

For those who require positions with higher accuracy, Differential Global Positioning Systems (DGPS) add a new element to GPS. This reference station receives

The reliability of the Japanese version of the ISMI scale was evaluated by testing the internal consistency of the items using Cronbach ’ s alpha, and by the scale ’ s test – re-

determination that fails to consider the concept of sovereign statehood and the society of states invites theoretical and pragmatic inconsistency and potential

Shape Memory Alloys (Sma's) Are Metals, Which Exhibit Two Very Unique Properties, Pseudo-Elasticity (An Almost Rubber-Like Flexibility, Under-Loading), And The

In the present study, we propose a novel type of ex- perimental population, specifically intended to exploit the power of inbred lines and saturated genetic maps, to

The categories include net sales, cost of goods sold, gross margin, selling and administrative expenses (or operating expense) and net profit.. These are categories

When data stored in the body of an email is read back from a server, the time taken for the transfer is equal to or less than that of the time taken when the data is stored in

“Fundamental Privacy Rights within a Shared Vision for Perimeter Security and Economic Competitiveness: Submission by the Office of the Privacy Commissioner of Canada to the