You are on page 1of 5

A SELF-ORGANIZING MULTI-CHANNEL MEDIUM ACCESS CONTROL (SMMAC) PROTOCOL FOR WIRELESS SENSOR

NETWORKS
Kae-Hsiang Kwong, Tsung-Ta Wu, Craig Michie, Ivan Andonovic
Institute for Communications and Signal Processing Department of EEE, Royal College Building, 204 George Street, Glasgow, United Kingdom

[kwong/ twu/c.michie/i.andonovic@eee.strath.ac.uk]

Abstract-This paper proposes a novel multi-channel medium access control (MAC) protocol designed for wireless sensor networks (WSN). Although multiple channel radio communication protocols are well understood in ad hoc wireless and other radio networks, most are not applicable in WSN as sensor nodes which are constrained by low cost and a primitive physical layer.
The proposed MAC allows data transmissions take place on multiple channels simultaneously within a single or multiple overlapped radio zones. This MAC applies a light weight, twotier handshaking algorithm permitting sensor nodes to negotiate a non-occupied channel dynamically. Experimental results show that SMMAC can improve the network performance compared with traditional single channel MAC.

A multiple channel MAC protocol, referred to as 'Selforganizing Multi-channel Medium Access Control (SMMAC)' is proposed. SMMAC increases network capacity, reduces packet loss and delay by using a number of radio channels simultaneously allowing information being transmitted in parallel across the network. The remainder of the paper is structured as follows: Section 2 firstly reviews a few representative multi-channel MACs currently used in wireless networks. A description of SMMAC is provided in Section 3. Section 4 discusses the experimentation results obtained from MICAz test bed. Finally the paper is concluded in Section 5.
II. RELATED RESEARCH

Keywords- WSN, medium access control, multiple channels, and MANET


I.

INTRODUCTION

Wireless sensor networks have evolved from simple monitoring/data acquisition applications to multi purpose data hungry applications. These new applications place higher demands on the networks; for example [1] has proposed a new hybrid network structure which integrates WSN into a wired network. In [2] the WSN design principle has migrated from low data volume to information-rich data applications in which data volume has increased sharply. The design issues of such applications have been discussed in other studies [3-4]. In general, a new trend can be observed in which applications in WSN are evolving from traditional low data rate monitoring to high data volume multi-purposes applications. This paper addresses the challenges associated with implementing a MAC as the bandwidth demand of WSN evolves from conventional low data applications into data hungry sensor network applications (i.e. video surveillance in battle field scenarios, or multiple sensors installed into a node sharing single radio interface).

A significant body of research has been reported [6-12] on multiple channel protocols in wireless networks. One noteworthy protocol is the 'Multi-channel MAC (MMAC)' [9] built on top of the IEEE 802.11 power-saving mechanism. MMAC is a time-slotted protocol in which bandwidth is divided into time slots comprising beacon period, ATIM window and data window. Nodes are configured to tune into a predefined signalling channel during the ATIM window, in so doing informing all nodes about all future transmissions and their corresponding channel allocations. Hence a more informed decision regarding their own channel allocation can be reached. MMAC requires network-wide temporal synchronization; although time synchronization is achievable in the WSN environment of low specification sensors nodes, nonetheless, frequent synchronization introduces an additional power overhead.

The 'Dual Busy Tone Multiple Access (DBTMA)' protocol [8] equips each node with multiple radio interfaces and two extra bands on top of the data channel. These 'extra' channels are not used for data transmission but are primarily used for resolving RTS/CTS collisions.

III.

THE SELF-ORGANIZING MULTI-CHANNEL MEDIUM ACCESS CONTROL (SMMAC) PROTOCOL

In this section, signalling packets i.e. Request Tune to Channel (RTC) and Channel to Send (CTS) are first introduced. A detailed description of SMMAC is then followed in which a two-tier procedure is defined allowing senders and receivers to agree on a non-occupied channel for transmission in a promptly manner. A. Signalling packetformat The packet structures of RTC and CTS are illustrated in Figure 1. These packets form a crucial part of SMMAC. They are used by senders and receivers during the channel negotiation process and they also eliminate issues caused by hidden terminal. The RTC packet is released by a sender to a receiver initialling a new session of packet transmissions. The RTS packet besides containing a source and a destination addresses which is used to identify the sender and the receiver. CCL and FragCount fields also present. In the CCL the sender describes the channel usage and informs the receiver which channels are currently free. On the other hand, the FragCount is used to inform the receiver the number of packets will be transmitted. The CTS packet is generated by the receiver in response to a RTC. A CTS packet only contains three fields: source address, destination address, and CCL field. The usage of source and destination addresses is apparent. The CCL filed is use to notify the sender the newly selected channel for communication.
CCL
Source address
Destination address
1 0 1 1 1 0 0 0

convenience represent channel "busy" and "free" respectively. For example, Figure 1 shows that the sender has detected that channel 1, 3, 4 and 5 are free and the receiver has selected channel 3 for transmission.

C. Protocol Description of SMMAC To aid in the understanding of SMMAC, 4 nodes (Node A D) and 3 channels (ch-1 signalling channel, ch-2 and ch-3 data channels) are used in the following, simplified example. Power management is not considered; all nodes are set to be always powered on. Even when nodes are idle, they will constantly monitor the signalling channel. Node A is defined as sender and Node B, the receiver. Both are required to negotiate a new channel for transmission. Node C and D have an established communication in ch-2. The network architecture is illustrated in Figure 2 where two radio effective zones are presented.

Node A

Node B

Node C

Node D'

RTC [ch-2, ch-3]

CTS
:

Ich-3]
ch-2 :in

ACK 11111111111111111111111111111111............................................................................................................. .................................................

Communications

FragCount
1 byte

2 bytes

bytes
(a)

1 byte

*
*-.--.....-..-..-..-..-...

~~Radio Zone 2

CCL
0 0 1 0 0 0 0 0
1

Source address Destination address


2 bytes 2 bytes

Figure 2. Illustration of SMMAC protocol.

byte

(b)
Figure 1. (a) RTC packet format; (b) CTS packet format.

B.

Candidate Channel List Each node maintains a data structure called the Candidate Channel List (CCL) indicates which data channel is available. The CCL of each node is updated in a just-in-time manner. For instance, when a sender wants to initiate a channel negotiation process, it will scan through all data channels and record the status of each channel. This just-in-time CCL update not only allows each node to maintain an accurate CCL, it can also reduce the power required to constantly monitor the radio channels. The presented protocol uses binary state to represent the channel condition. The designations "O" and "1" simply for

The SMMAC algorithm for negotiating a new channel is as follows: 1. Node A scans through all available data channels. From Node A's perspective, all data channels are available. Hence Node A will generate a RTC indicating that both ch-2 and ch-3 as free channels. This is then released in the signalling channel. Here, it is assumed that the RTC has been transmitted successfully. 2. Once Node B receives the RTC, it knows that Node A is requesting to establish a communication channel from the source and destination address in the RTC. Node B will select ch-3 to be data channel since ch-2 is occupied. It will generate a CTS selecting ch-3 and reply to Node A in the signalling channel.

(a)

(b)
Figure 3. (a) SMMAC algorithm for sender; (b) SMMAC algorithm for receiver.

3. After receiving CTS released by Node B, Node A will reply Node B with an ACK and then switch immediately to ch-3 and begin transmission. 4. If Node B received an ACK in return, it means that the CTS has been received by Node A successfully and it will switch to ch-3 and be ready to receive a packet from Node A. Otherwise, after waiting for a predefined time period, it determines that the CTS is lost or has collided and it will perform a CTS retransmission. 5. Both Node A and B will switch back to signalling channel as soon as the communication process is completed.
IV. EXPERIMENTATION

350
300 250 a

SMMAC-CH2-Fragl

BMAC
v;7

SMMAC-CH4-Fragl SMMAC-CH2-Frag2

SMMAC-CH4-Frag2

200
150
........

50 z

50

100

150

200

250

300

350

Offered traffic load (Kbps)

Figure 4. Network throughput comparison between BMAC and SMMAC.

In the following experiments 30 MICAz nodes [14] (15 sender-receiver pairs) were used. These nodes were placed within transmission range of each other and formed a single radio effective zone. The data packet size was set to be 110 bytes and a Poisson distribution used to generate data stream. The receivers utilised a counter to record the number of packets successfully received. At the end of the experiment, the receivers reported on the counter value to a base station. The results obtained from the experiments are shown in the following diagrams. In the legend of these diagrams, 'BMAC' is used to denote the result of single channel B-MAC protocol [5]. Whereas, 'SMMAC-CHx-Fragy' indicates the results of SMMAC protocol; x represents the number of data channels used in the experiment and y shows the number of packets being released by the sender in each transmission session. Figure 4 provides a comparison of network throughput between B-MAC and SMMAC with different number of available data channels. The x-axis displays the offered data traffic load which is the aggregated traffic load from all 15 senders. In these experiments, B-MAC has saturated when offered load reaches 158 Kbps and the network throughput starts degrading beyond this point as high collision rate has deferred the senders from sending. The results of SMMAC have clearly shown that network throughput is, in general, improved as the total network capacity is multiplied by introducing extra data channels. The experiments also indicate that results obtained from 4-data channel SMMAC are close to 2-data channel SMMAC when only one packet is transmitted by the senders after each RTC/CTS exchange. Both of these systems become saturated when offered traffic load reaches 250 Kbps. This is an interesting finding as 4-data channel SMMAC is expected to have a far better performance given the fact that its total network capacity is 500 Kbps more that 2-data channel SMMAC. This phenomenon is caused by congestion in the signalling channel. When the offered traffic load increases the amount of RTC/CTS generated by senders and receivers will increase linearly and all of these packets are transmitted in the signalling channel. This problem can be resolved by allowing the senders to transmit more than one packet in each session. In the SMMAC protocol, the sender needs to inform the receiver the amount of packets will be transmitted during RTC/CTS exchange, to do so, the sender needs to fill in the number of packets for transmission in FragCount.

0.6

BMAC

SMMAC-CH2-Fragl SMMAC-CH4-Fragl 0.5 SMMAC-CH2-Frag2 SMMAC-CH4-Frag2


0.4

0n 0.3cn
02 0.1

0t 50

100

150

200

250

300

Offered traffic load (Kbps)

Figure 5. Packet losses of MAC protocols with different offered traffic load.

800 SMMAC-CH2-Frag5 SMMAC-CH4-Frag5 700 600 Q o 500

SMMAC-CH6-Frag5

400

200 100

00 300

400

500

600

700

800

Offered traffic load (Kbps)

Figure 6. Throughput of SMMAC protocols with different numbers of available data channels in high offered traffic load.

Figure 5 shows the amount of packet loss of above experiments. Once again, for the SMMAC systems, when only one packet is transmitted in each session, the network is limited by congestion in the signalling channel causing substantial losses. The figure also shows that the packet loss can be reduced by allowing more than one packet being transmitted in each session. Figure 6 shows the results of SMAAC with 2, 4, and 6 data channels. In these experiments, 5 packets are transmitted by senders in each successful exchange of RTC/CTS. The results

indicate that SMMAC with its light-weight protocol scaled up to support 6 data channels or even more.
V. CONCLUSIONS

can

be

A Self-Organizing Multi-Channel MAC (SMMAC) is proposed, and further developed and tested on MICAz hardware platform. Tinyos was selected and used for software development. Experimental results have shown the potential of this light-weight protocol to be one of the top candidates for migrating WSN into futuristic information rich applications. The next steps of this research will include enhanced SMMAC with mobility support, and in depth studies of SMMAC performance in a multi-hop relay environment. These future studies will bring the R&D of WSN closer to implementing a high data rate ad hoc network on top of low cost/specification hardware platform.
REFERENCES Rui Teng, "RBR: A Region-Based Routing Protocol for Mobile Nodes in Hybrid Ad Hoc Networks", IEEE Communications Magazine, Vol.44, No. 11, Nov 2006. Yonghe Liu, Sajal K. Das, "Information-intensive Wireless Sensor Networks: Potential and Challenges", IEEE Communications Magazine, Vol.44, No. 11, Nov 2006. P. M. Ruiz, F. J. Ros, and A. G. Skarmeta, "Internet Connectivity for Mobile Ad Hoc Networks: Solutions and Challenges" IEEE Commun. Mag., Oct. 2005.

[1] [2] [3]

Y. Sun, E. M. B. Royer, and C. Perkins. "Internet Connectivity for Ad Hoc Mobile Networks," Int'l. J. Wireless Info. Networks, vol. 9, Apr. 2002. [5] J. Polastre, J. Hill, D. Culler: Versatile Low Power Media Access for Wireless SensorNetworks. ACM SenSys, 2004 [6] Jungmin So, Nitin Vaidya, "Multi-channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden Terminals Using A single Transceiver", MobiHoc 2004. [7] Dong Zheng, Junshan Zhang, "Protocol Design and Throughput Analysis of Frequency-Agile Multi-Channel Medium Access Control", IEEE Transactions on Wireless Communications, VOL. 5, NO. 10, Oct 2006. [8] Jingpu Shi, Theodoros Salonidis, Edward W. Knightly "Starvation Mitigation Throughput Multi-Channel Coordination in CSMA Multi-hop Wireless Networks", Dept. of Electrical and Computer Engineering, Rice University Houston. [9] Ritesh Maheshwari, Himanshu Gupta and Samir R. Das, "Multi-channel MAC Protocols for Wireless Networks", Dept. of Computer Science, Stony Brook University. [10] Jaehyuk Choi, Joon Yoo, Sunghyun Choi, Chongkwon Kim, "EBA: An Enhancement of the IEEE 802.11 DCF via Distributed Reservation", IEEE Transactions on Mobile Computing, Vol. 4, No. 4, Jul/Aug 2005. [11] Zygmunt J. Haas, Jing Deng, "Dual Busy Tone Multiple Access (DBTMA) A Multiple Access Control Scheme for Ad Hoc Networks", IEEE Transactions of Communications Vol.50, No. 6, Jun 2002. [12] Gang Zhou, Chengdu Huang, Ting Yan, Tian He, John A. Stankovic, Tarek F. Abdelzaher "MMSN: Multi-Frequency Media Access Control for Wireless Sensor Networks", Dept. of Computer Science, University of Virginia. [13] TinyOS website, http://www.tinyos.net [14] Crossbow Technology Inc., http://www.xbow.com

[4]

You might also like