REAL-TIME MULTICAST ROUTING IN WIRELESS

SENSOR NETWORKS

Hosung Park, Jeongcheol Lee, Seungmin Oh, Yongbin Yim, and Sang-Ha Kim

Department of Computer Engineering, Chungnam National University, Daejeon, Republic of Korea

Keywords: Wireless Sensor Networks, Multicasting, Real-time Communication.

Abstract: Real-time data dissemination to a multicast group is data delivery to each member in the multicast group

within the desired time deadline. The hardest aspect of this mission is to enforce the real-time constraint in

the communication between a source and the furthest member since an end-to-end delay is proportional to a

physical distance in wireless sensor networks. We call it the critical distance. The critical distance should be

most important constraint for real-time multicasting. That is, the delivery distance from a source to each

member should not be longer than the critical distance even by any reason. However, since the traditional

multicast protocols lay the strong emphasis only on the overall communication cost rather than delivery

distance to each member, they may violate the real-time constraint related to the critical distance. In this

letter, we propose a novel multicast protocol for real-time data dissemination.

1 INTRODUCTION

Many sensor network applications, such as

battlefield surveillance and fire alert, are designed to

interact between a fast changing event (He, 2005)

and multiple destinations (Akyildiz, 2010) in the real

world. In these applications, it is often necessary for

multicast protocol to meet real-time constraint. Real-

time data delivery to a multicast group, i.e. real-time

multicasting, may be defined as data dissemination

to each member in the multicast group within the

desired time deadline.

In wireless sensor networks, unlike legacy

networks, the hop count increases as the physical

distance packet travels increases since the radio

range of each node is bounded. The end-to-end delay

is also increases due to the increase of hop count. In

other words, the end-to-end delay is proportional to

the physical distance packet travels (He, 2005).

Therefore, the hardest part of real-time multicasting

is to enforce the real-time constraint from a source

node to the furthest destination in the multicast

group. In view of this, we refer to linear distance

between the source node and the furthest destination

as a critical distance. The critical distance must be

most important constraint for real-time multicasting.

In other words, the delivery distance from the source

node to each destination on the multicast tree must

not be longer than the critical distance even by any

reason.

In wireless sensor networks, multicast protocols

(Wu, 2007), (Sanchez, 2007) are proposed to deliver

data to multiple destinations. They try to achieve a

single goal, constructing cost-efficient multicast tree

that minimizes the total delivery path from the

source node to all destinations. Since the traditional

multicast protocols lay the strong emphasis only on

communication cost rather than each delivery

distance, they may violate the real-time constraint

related to the critical distance.

In this letter, we propose a novel Multicast

Protocol for Real-time Data dissemination (MPRD).

MPRD constructs multicast tree considering the

critical distance and disseminates data by

spatiotemporal approach (He, 2005) via the

multicast tree to satisfy the real-time constraint.

2 REAL-TIME MULTICASTING

The problem we are facing can be described as

follows: a data generated by a source node is

delivered to multiple destinations with real-time

constraint. In other words, each of the destinations

wants to receive the data within the desired time

deadline. In this section, we present MPRD for

achieving this goal in detail.

407

Park H., Lee J., Oh S., Yim Y. and Kim S..

REAL-TIME MULTICAST ROUTING IN WIRELESS SENSOR NETWORKS.

DOI: 10.5220/0003905604070410

In Proceedings of the 2nd International Conference on Pervasive Embedded Computing and Communication Systems (PECCS-2012), pages 407-410

ISBN: 978-989-8565-00-6

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

Source

Figure 1: Real-time multicast tree construction.

2.1 System Model

MPRD, like other geographic multicast protocols

(Wu, 2007), (Sanchez, 2007), employs geographic

greedy forwarding as the underlying routing

protocol. In wireless sensor networks, geographic

multicast protocols are considered efficient, because

they use only location information without the

topology information of a whole sensor field. This

paper is focused on multicast tree construction and

data dissemination. We therefore assume throughout

this paper that the source node is aware of locations

of destinations. To know that information, the source

node can employ destination location service

schemes such as (Liu, 2006).

2.2 Real-time Multicast Tree

Construction

When the source node generates real-time data, the

source node constructs virtual multicast tree rooted

at itself for real-time data dissemination to multiple

destinations. On the multicast tree, each distance of

path from the source node to all destinations has to

be shorter than the critical distance. Note that the

multicast tree is the virtual structure calculated by

the source node instead of real structure constructed

in practice.

Given a set of the destinations, MPRD connects

the source node with the furthest destination in the

linear path as the base line; then the path splits into

branches toward other destinations. Figure 1 shows

the process. MPRD draws a straight line from the

source node to the D5, the furthest destination, as a

base line. Other destinations, D1-D3 are connected

with the base line perpendicularly since each line is

shortest path to the base line.

However, some destinations may have longer

path from the source node than the critical distance.

Destinations located in the shaded area in figure 1

Source

Sector 1

Sector 2

Sector 3

Figure 2: Dividing sectors.

belong to this group. The D4 therefore needs to be

connected with the base line in other way. When

each destination is connected with the base line,

MPRD checks a condition regarding the critical

distance. If the path from the source is longer than

the critical distance (if the destination is located in

the shaded area), MPRD select a branch point on the

base line closer to the source node in order to make

the path from the source node to the D5 to be shorter

than or equal to the critical distance. In figure 1,

MPRD determines the branch point B4 which makes

the distance from the B4 to the D4 and the distance

from the B4 to the D5 to be equal; accordingly, the

distance of the path between the source node and the

D4 is equal to the critical distance.

Destinations in the opposite area of the first base

line cannot be perpendicularly connected with the

first base line. MPRD therefore divides the network

into two or three sectors based on locations of the

source node and the furthest destinations in each

sectors and constructs paths separately. In figure 2,

sector 1 is divided by the L1, orthogonal dotted line

of the base line from the source node to the D1. A

sub-tree rooted at the source node is constructed by

connecting destinations in sector 1 with the base

line. Sector 2 is divided by the L2, orthogonal dotted

line of another base line from the source node to the

D2 in residual area. The D2 is the furthest

destination among all destinations but destinations in

sector 1. Another sub-tree rooted at the source node

is constructed by connecting destinations in sector 2

with the base line the source node to the D2. The

third sub-tree is constructed by the same way in

sector 3, residual area except sector 1 and sector 2.

MPRD has low computation complexity due to

the simple method that each destination is

perpendicularly connected with the base line after

checking just one condition for the critical distance.

Dividing network is also useful since it make MPRD

be able to apply the simple method to each sector.

PECCS 2012 - International Conference on Pervasive and Embedded Computing and Communication Systems

408

2.3 Data Dissemination

After the source node calculates the virtual real-time

multicast tree, the source node sends real-time data

to all destinations via the multicast tree.

In wireless sensor networks, to satisfy the desired

time deadline, real-time routing scheme proposed in

(He, 2005) utilize a spatiotemporal approach by

which data are delivered with a delivery speed

obtained from both the desired time deadline and the

distance from a source to a destination. Every

immediate node elects the next-hop node that is one

of one-hop neighbor nodes closer to the destination

than itself and has a relay speed larger than the

delivery speed. Hence, by relaying data with a faster

speed than the delivery speed per every hop, data

can be delivered on the desired time deadline. The

spatiotemporal approach can be applied due to the

property that data delivery delay is dependent on the

distance a packet travels in wireless sensor networks.

We apply the spatiotemporal approach in order to

disseminate real-time data.

After the source node calculates the virtual real-

time multicast tree, the source node encapsulates

calculated tree information and delivery speed in the

data packet and disseminates it along the calculated

multicast tree. The tree information is the locations

of destinations and branch points, and the delivery

speed is calculated with the desired time deadline

and the distance of the base line. Note that the tree

information and the delivery speeds are different at

each sector.

In the data dissemination process, each node

forwarding the packet selects a next-hop node

among one-hop neighbors that satisfy the condition

of the spatiotemporal approach. In figure 1, the data

packet is delivered toward the branch point B1 via

the spatiotemporal approach. When a sensor node

1 4 7 10131619222528

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Deadline Success Ratio

Simulation Round

GMP

GMR

MPRD

Figure 3: Deadline success ratio at each simulation round.

receives the data, if the B1 is located in radio range

of the node and the node has no neighbor which is

more closely located to the B1, the node elects itself

as a branch node for the branch point B1. The

branch node selects two next-hop nodes toward the

D1 and the B2. The location of the D1 is

encapsulated in packet toward the D1 and tree

information except the locations of the B1 and the

D1 is encapsulated in packet toward the B2. The

delivery speed of the packet toward the D1, of

course, could be recalculated by the distance on the

path from the source node and the D1. This process

is repeated until all destinations receive the data

packet.

3 PERFORMANCE EVALUATION

In this section, we present simulation results to

evaluate performance of MPRD. The purpose of

simulations is verification that MPRD has higher

success ratio of real-time data dissemination than

GMP (Wu, 2007) and GMR (Sanchez, 2007)

applying the spatiotemporal approach. We also

evaluate energy efficiency: one of the most

important issues in wireless sensor networks.

3.1 Simulation Environments

and Metric

We simulate MPRD on QualNet simulator. The

simulation network space is 1000 X 1000 m2. The

number of nodes is 5000 and the number of

destinations is 20. Transmission range of each node

is 30 m. Node and destination placement follows

random deployment. The source node is located at

the center of network and sends 20 data packets at

each round. The desired time deadline for real-time

constraint is 1 second and delay of each node is

randomly set between 0.03-0.1 seconds.

Deadline success ratio is average ratio of the

number of destinations that receive data packets

within the time deadline to the number of

destinations.

3.2 Simulation Results

Figure 3 shows the deadline success ratio at each

simulation round. We randomly deploy sensor nodes

and destinations at each simulation round. The

deadline success ratio of MPRD is distributed in

closer to 1 than other protocols. In other words,

MPRD has higher probability of satisfying the real-

time constraint. In GMP and GMR, the data is

delivered to some destinations by longer path than

the critical distance. MPRD however delivers data to

REAL-TIME MULTICAST ROUTING IN WIRELESS SENSOR NETWORKS

409

10 12 14 16 18 20 22 24 26 28 30

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Deadline Succes Ratio

The Number of Destinations

GMP

GMR

MPRD

Figure 4: Deadline success ratio in respect of the number

of destinations.

all destinations along paths shorter than or equal to

the critical distance. Since end-to-end delay is

dependent on the physical distance a packet travels

in wireless sensor networks, MPRD has higher

chance of satisfying the real-time constraint.

Figure 4 shows the deadline success ratio in

respect of the number of destinations. As the number

of destinations increases, the deadline success ratio

of GMP decreases. In GMP, more paths from the

source node to destinations are longer than the

critical distance as the number of destinations

increase, since the multicast tree becomes more

complicated. In addition to the reason, GMR spends

more time in order to forward data as the number of

destinations increases, since computation complexity

of every forwarding node increases exponentially.

MPRD however is unaffected by the number of

destinations, since MPRD delivers data to all

destinations along paths shorter than or equal to the

critical distance.

0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400

0.0

0.5

1.0

1.5

2.0

Average Energy Consumption (J/node)

Simulation time (sec)

GMP

GMR

MPRD

Figure 5: Energy consumption on the simulation time.

Figure 5 shows total energy consumption on the

simulation time. Since three multicast protocols has

no signaling overhead, energy consumption is

dependent on total hop counts for data

dissemination. Total tree length of MPRD is little

longer than that of GMP and GMR in order to make

each path shorter than the critical distance. Since the

distance packet travels is proportional to the hop

count as mentioned above, the total number of

packet transmission in MPRD is more than that in

GMP and GMR. Therefore total energy consumption

of MPRD is little higher than GMP and GMR.

4 CONCLUSIONS

We propose a new multicast protocol for real-time

data dissemination in wireless sensor networks. To

deliver data to multiple destinations with real-time

constraint, MPRD utilizes the property that data

delivery delay is dependent on the distance a packet

travels in wireless sensor networks. We refer to

linear distance between the source node and the

furthest destination as a critical distance. MPRD

constructs multicast tree that every delivery distance

from the source to each destination on the multicast

tree is shorter than the critical distance and apply the

spatiotemporal approach. Simulation results show

that MPRD has better performance than traditional

multicast protocols in terms of real-time data

dissemination.

REFERENCES

He, T., Stankovic, J. A., Abdelzaher, T. F., and Lu, C.,

2005, A spatiotemporal communication protocol for

wireless sensor networks. IEEE Transaction on

Parallel and Distributed Systems, vol. 16, no. 10, pp.

995-1006.

Akyildiz, I. F. and Vuran, M. C., 2010. Wireless Sensor

Networks. John Wiley and Sons, Ltd.

Wu, S. and Candan, K. S., 2007. Demand-scalable

geographic multicasting in wireless sensor networks.

Computer Communications, vol. 30, pp. 2931-2953.

Sanchez, J., Ruiz, P., Liu, J., and Stojmenovic, I., 2007.

Bandwidth-Efficient Geographic Multicast Routing

for Wireless Sensor Networks. IEEE Sensors Journal,

vol.7, no. 5, pp. 627-636.

Liu, D., Stojmenovic, I., and Jia, X., 2006. A scalable

quorum based location service in ad hoc and sensor

networks. In MASS’06, IEEE International

Conference on Mobile Ad-Hoc and Sensor System.

PECCS 2012 - International Conference on Pervasive and Embedded Computing and Communication Systems

410