Chapter 2 Related Work
2.2 Comparison of multi-channel MAC protocols
The approaches mentioned in the previous section are qualitatively compared in terms of protocol type, number of rendezvous, number of radios, basic idea, backoff mechanism and control overhead, as shown in
protocols
Table I. The proposed EMC-MAC is also included in the table, which will be described in Section 3. In Table I, the protocol type indicates the operation type of these multi-channel MAC protocols as classified in [8]. The number of rendezvous indicates the number of control channels that can be used for all the devices. The number of radios indicates how many transceivers equipped on each device. The basic idea indicates the main idea of each protocol. The backoff mechanism indicates how to backoff when channel contention occurred. The last metric, control overhead, indicates that the amount of exchanged control messages.
The first four approaches in Table I are contention-based multi-channel protocols. Most contention-based multi-channel MAC protocols are not suitable for delay-sensitive WMSN
Table I. Comparison of existing multi-channel MAC protocols with proposed EMC-MAC.
No. radios Basic idea Backoff mechanism
Control
Each node uses one radio to monitor the control channel together and stops upon agreement for
Split phase Multiple
N
Split phase Multiple
N
6
because each packet has to contend for medium access and the delay for data delivery could be potentially unbounded. The amount of time required to resolve collisions depends on the load condition of the network, which makes it very difficult to guarantee a bounded delay [4].
In contrast, COM-MAC [4] and the proposed EMC-MAC use the cluster head to coordinate all its cluster members to reduce control messages and let multiple nodes transmit their multimedia data at the same time through multiple channels. In the request phase, both protocols use the contention-based protocol while in the transmission phase, they use the contention free protocol to transmit multimedia data. Therefore, both protocols can guarantee a bounded delay. In the next section, we will further compare the main differences between the proposed EMC-MAC and COM-MAC [4].
2.3
COM-MAC [4] bases traffic load to determine to use a contention free or contention based protocol at the control channel. However, it did not propose any mechanism to decide whether a node needs to send messages or not and it uses a static request session and the duration to transmit a request to define the backoff interval. It may cause some nodes unable to transmit its request at the current round; thus, the medium access delay increases and the network throughput decreases.
Comparison of proposed EMC-MAC and COM-MAC
The above problems motivate us to design a more efficient multichannel MAC protocol for cluster based WMSNs. We propose an energy efficient mechanism to choose a node to send its multimedia data. In addition, an enhanced sliding contention window [7] is proposed to effectively use the control channel. Table II summarizes the main differences between COM-MAC and the proposed EMC-MAC.
Table II. Comparison of proposed EMC-MAC and COM-MAC.
Approach Backoff mechanism Energy efficiency
COM-MAC [4]
Cause The backoff mechanism
sometime gets a too large backoff interval which makes a node can’t transmit its request in the current round
All the nodes need to transmit multimedia to the associated cluster head
Effect It may increase the MAC delay and decrease the network throughput
Consume more energy and reduce the network life time
EMC-MAC (proposed)
Cause The backoff mechanism lets nodes with different request priorities to get different backoff intervals
When multiple nodes sense a same event, only the node has the highest energy level needs to send sensed data to the cluster head
Effect Decrease the MAC delay and increase the network throughput
Save more energy and prolong network life time
8
Chapter 3
Design of Proposed Efficient Multi-channel MAC Protocol
3.1 Network architecture
As shown in Figure 1, a WMSN consists of several powerful nodes (cluster heads) located at the center of different monitoring areas, a number of stationary multimedia sensor nodes surrounding around a related cluster heads, and a remote sink which stores multimedia contents for later retrieval. Each cluster head can communicate with the sink directly using an out-of-band channel. However, if one hop communication to the sink is unavailable, multi-hop routing can also be employed [4].
Figure 1: Network architecture.
We apply this architecture in Figure 1 to an example environment, such as a hospital, to illustrate its possible application scenarios. In Figure 2, each floor forms a cluster and the cluster head is placed at the nursing station which is in the center of the floor. Each ward of the hospital is deployed with two multimedia sensor nodes for fault tolerance. A node within a cluster is either active or inactive. As shown in Figure 3, nodes which need to send multimedia data to the cluster head are active, while the others are inactive.
Nursing Station
201
202
203 208
209
210
215
216
217 222
223
224
Cluster head
Cluster member
Figure 2: Deployment of sensor nodes in a hospital.
10
3.2 Hardware assumptions
We make the following assumptions regarding the configuration of a WMSN [4].
1) There are N different channels available and all channels have the same bandwidth.
2) Each multimedia sensor node is equipped with a single half-duplex transceiver, which means a sensor node is unable to transmit and receive data at the same time.
3) A multimedia sensor node can only transmit or receive on one channel at a time. It is able to switch among channels dynamically. The channel switching time is less than 224 µs according to [6].
4) Each cluster head is equipped with N half-duplex transceivers, which means a cluster head can transmit or receive on N channels simultaneously. In addition, each cluster head has sufficient power supply and better processing ability.
5) All the cluster members are synchronized by its related cluster head and each cluster member can communicate with its cluster head directly.
Cluster head
Inactive nodes Active nodes
Figure 3: States of cluster members.
3.3 Proposed EMC-MAC protocol
In this section, the proposed efficient multi-channel medium access control (EMC-MAC) protocol for WMSNs is described. We assume the clustering process has been completed by some existing clustering protocol and all the multimedia nodes have joined the nearest cluster head [4]. Within a cluster, the control channel assignment phase will be executed at first, then the following three phases, request phase, scheduling phase, and transmission phase, will be executed sequentially in a round, as shown in Figure 4. The length of a round would vary according to network traffic load. All these phases will be described in the next subsection.
At first, each node receives a pre-allocated control channel assignment from its cluster head. Figure 5 shows the flowchart of node behavior in each round. Once a node sensed an event, the node starts to send its request message through a pre-allocated control channel. If control channel contention occurred, the contention nodes start to execute a backoff algorithm based on an enhanced sliding contention window [7], which will be discussed in Section 3.3.2.
After the request message transmitted successfully, the node waits for a scheduling message
Request phase Scheduling phase Transmission phase
Round 1 Round 2
Round 3
Control channel assignment phase
Figure 4: Round structure.
12
from the cluster head and checks whether it is an active node or not. If yes, the node starts to transmit its sensed data to the cluster head in a given timeslot on an assigned radio channel.
3.3.1 Control channel assignment phase
In this phase, all the N channels can be assigned to different nodes as their control channels. To avoid possible channel contention, different channels will be assigned to adjacent nodes because these nodes are close to one another that tend to sense the same event and transmit sensed data at the same time.
3.3.2 Request phase
After the control channel assignment phase, any node that sensed an event can send a REQ message to the cluster head. This message contains the node ID, size of multimedia data to be transmitted, energy level, and priority of the data, which is decided by the emergency degree of data, for example, high priority for data from an intensive care unit. Each node
Sensed event
Send its sensed data to an assigned radio
channel and
Figure 5: Flowchart of node behavior in each round.
sends an REQ message through the pre-allocated control channel in a contention-based manner. In order to avoid possible collisions, an enhanced MILD based sliding contention window mechanism is employed in each sensor node [7]. This mechanism lets a request message with a higher priority get a contention window that results in a smaller backoff interval, whereas a request message with a lower priority receive a contention window that results in a longer back off interval. In this way, the channel contention probability may decrease and the throughput may increase. The backoff interval for node i is initialized as follows. CWLB[i]=CWmin is the lower bound of the backoff interval. CWUB[i]=CWmin+2× SF is the upper bound of the backoff interval, where SF is a sliding factor. The parameter settings of MILD sliding contention window are shown in Table III. We can let a request node with high priority has a small backoff interval which is (0 ~ 4), a request node with medium priority has a medium backoff interval (3 ~ 11), and a request node with low priority has a large backoff interval (7 ~ 23). That is, we set parameters in a way to reduce channel contention and the waiting time required to transmit a request. In order to determine an appropriate interval for the request phase, we executed the EMC-MAC 100 rounds for each cluster size. Then choose an average of these 100 rounds’ request intervals as the final request interval.
Table III. Parameter settings of the enhanced MILD based sliding contention window.
Traffic priority SF (time slots)
14
3.3.3 Scheduling phase
After the request phase, the cluster head start to select which nodes need to send sensed multimedia data and becomes active nodes according to the received REQ messages. We use the energy level of each node to select active nodes, as illustrated in Figure 6. In Figure 6, nodes 1 and 2 monitor the same ward, and nodes 3 and 4 monitor the same ward. If some emergent event happened, all nodes in the corresponding ward will send REQ messages to the cluster head. Assume node 1’s energy level is higher than node 2’s energy level and node 3’s energy level is higher than node 4’s energy level. The cluster head will choose nodes 1 and 3 to send the sensed multimedia data back. In this way, the number of nodes in a ward that need to send multimedia data back can be reduced and the lifetime of sensed nodes can be extended.
After a cluster head chooses active nodes for data transmission, the cluster head starts to schedule the data transmission of the active nodes. In order to achieve the QoS requirement, the active nodes with the same priority are grouped together. The group has the highest priority will send multimedia data first. In the same group, the order to transmit multimedia data is based on data transmission time. A node with the smallest data size will be assigned to
Node 1
Node 2
Node 3
Node 4
Figure 6: When multiple nodes sense the same event, only the node has the highest energy level needs to transmit multimedia data back.
a channel first. After the cluster head creates a schedule, it will broadcast this schedule to all the cluster members.
3.3.4 Transmission phase
After receiving a scheduling message from the cluster head, a cluster member will check if it is allowed to send back the sensed data. If yes, it becomes an active node and starts to send the sensed multimedia data on a specific radio channel and timeslot assigned by the cluster head; otherwise, it will ignore the scheduling message.
16
Chapter 4
Performance Evaluation
In this section, we first describe simulation setup and evaluation metrics. Then, we compare the proposed EMC-MAC with COM-MAC [4] in terms of the rounds till the first node death, network throughput, and MAC delay.
4.1 Simulation setup and evaluation metrics
We have developed an object oriented simulator using C++ for WMSN MAC protocols.
Related simulation parameters are shown in Table IV
The energy consumption model [16] has adopted and is described as follows. ETx( dk, )
denotes the energy consumption of a node to transmit a k-bit message over a distance d due to the energy consumption of the transmitter circuitry ETx−elec(k) and the transmitter amplifier
[20].
d) (k,
ETx−amp Thus, ETx( dk, ) is given by:
Table IV. Simulation parameters [4, 16].
Parameter Value
Number of channels 3
Channel bandwidth 250 Kbps
Packet size 525 bytes
Transmission range 10 m
Packet arrival rate Randomly choose from [0,10]
Initial node energy level 6 J
Eelectransmitter circuitry dissipation per bit 50 nJ/bit εamptransmit amplifier dissipation per bit per square meter 100 pJ/bit/m2
(1)
Similarly, ERx(k)represents the energy consumption of a node to receive a k-bit message due to the energy consumption of the receiver circuitry ERx−elec(k).ERx(k)is given by:
4.1.1 The rounds till the first node death
We execute both the protocols, COM-MAC and EMC-MAC, until the first node death and record total rounds passed when this node is out of energy. This metric, the round till the first node death (FND), indicates the energy balancing capability of a protocol.
4.1.2 Throughput
We use successfully received data packets in the cluster head divided by the total elapsed time of the network as throughput (Mbps), which is defined as follows:
4.1.3 Average MAC delay
Average MAC delay is defined as the elapsed time between the first attemp time to send a REQ message to the time actually to send a REQ message, which is defined as follows:
Throughput =
18
4.2 Comparison between EMC-MAC and COM-MAC
Figure 7 shows the rounds till the first node death under different cluster sizes, where the cluster size is the total number of sensor nodes in a cluster. Therefore, the proposed EMC-MAC protocol achieves better consumed energy balancing than COM-MAC, since we base on the energy level of each node to select active nodes for multimedia data transmission.
The round till the first node death of EMC-MAC is 72.24% larger than that of COM-MAC. It also implies that EMC-MAC is more energy efficient than COM-MAC.
Davg=
For comparison of the throughput of EMC-MAC with that of COM-MAC under different cluster sizes was executed each protocols for 50 rounds. As shown in Figure 8, EMC-MAC is 23.72% better than COM-MAC in terms of throughput. This is because EMC-MAC has an enhanced sliding contention window that can effectively reduce the probability of channel contention. As a result, the time interval in the request phase can be reduced and a high throughput can be obtained.
Figure 7: The rounds till the first node death (FND) under different cluster sizes.
500 1000 1500 2000 2500 3000 3500
25 30 35 40 45
EMC-MAC (proposed) COM-MAC
Cluster size
FND
20
Figure 9 shows the comparison of the average MAC delay between EMC-MAC and COM-MAC. The average MAC delay increases as the cluster size increase for both protocols.
When the cluster size increases, channel contention increases. EMC-MAC is 44.59% better than COM-MAC in terms of average MAC delay. This is because EMC-MAC uses an enhanced sliding contention window in the backoff mechanism. The backoff mechanism let different priority nodes get different backoff intervals, so the channel contention probability can be reduced and nodes can send packets as soon as possible.
Figure 8: Throughput performance under different cluster sizes.
0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
25 30 35 40 45
EMC-MAC (proposed) COM-MAC
Cluster size
T hr o ug h pu t ( M bp s)
Figure 9: Average MAC delay performance under different cluster sizes.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
25 30 35 40 45
COM-MAC EMC-MAC (proposed)
Cluster size
A v er a g e M A C d elay( se c)
22
Chapter 5 Conclusion
5.1 Concluding remarks
In this thesis, we have presented an efficient cluster-based multi-channel MAC protocol for wireless multimedia sensor networks. We use the energy level of each node to select active nodes for multimedia data transmission and adjust the sliding contention window to effectively utilize control channels. Simulation results have shown that the proposed EMC-MAC’s throughput is 23.72% higher than COM-MAC’s, its average medium access delay is 44.59% lower than COM-MAC’s, and its rounds till the first node death is 72.24%
longer than COM-MAC’s. With the low MAC delay feature, our EMC-MAC protocol is feasible for applications of real time multimedia traffic sensing and transmitting, such as remote monitoring of hospital patients and fire spots.
5.2 Future work
A cluster head may be enabled with more functionalities, such as classifying received data according to their priorities to allow high priority data to be transmitted to the sink first, so that the proposed EMC-MAC can be more suitable for real-time multimedia applications, such as streaming video. In addition, we can integrate the proposed EMC-MAC protocol to an efficient cluster-based routing protocol to transmit multimedia data from source to sink efficiently.
Bibliography
[1] I. F. Akyildiz, T. Melodia, and K. R. Chowdry, ‘‘A Survey on Wireless Multimedia Sensor Networks,’’ Computer Networks (Elsevier) Journal, March 2007, pp.921 – 960.
[2] ‘‘XBOW MICA2 Mote Specifications,’’ [Online] Available: http://www.xbow.com/
[3] N. Saxena, A. Roy and J. Shin, “A QoS-based Energy-aware. MAC Protocol for Wireless Multimedia Sensor Networks
[4] C. Li, P. Wang, H. Chen, and M. Guizani, "A Cluster Based On-demand Multi-Channel MAC Protocol for Wireless Multimedia Sensor Networks," in Proc. IEEE International Communications Conf. ICC’08, May 2008, pp. 2371 – 2376.
,”in Proc. IEEE Vehicular Technology Conf. VTC’08, spring 2008, pp.183 –187.
[5] J. So and N. Vaidya, ‘‘Multi-channel mac for ad hoc networks: Handling multi-channel hidden terminals using a single transceiver,’’ in Proc. ACM, Roppongi Hills, Tokyo, Japan, May 2004, pp. 222-233.
[6] IEEE 802.11 Working Group, ‘‘Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications,’’ 1997.
[7] M. Bharatiya and S. Prasad, ‘‘MILD based sliding contention window mechanism for QoS in Wireless LANs,” in Proc. Wireless Communication and Sensor Networks, WCSN’08, 2008, pp. 105-110.
[8] Jeonghoon Mo, Hoi-Sheung Wilson So and Jean Walrand, ‘‘Comparison of Multi-Channel MAC Protocols,” IEEE Trans. Mobile Computing, vol. 7, no. 1, pp. 50-65, Jan. 2008.
[9] S. L. Wu, Y. Lin, Y. C. Tseng, and J. P. Sheu, “A New Multi-Channel MAC Protocol with On-Demand Channel Assignment for Mobile Ad Hoc Networks,” in Proc.
24
IEEE International Symposium
[10] S. L. Wu, C. Y. Lin, Y. C. Tseng, C. Y. Lin, and J. P. Sheu, “A Multi-Channel MAC protocol with Power Control for Multi-Hop Mobile Ad Hoc Networks,” The Computer Journal, vol. 45, no. 1, pp. 101-110, 2002.
on Parallel Architectures, Algorithms and Networks, ISPAN ’00, Dec. 2000, pp.232 - 237.
[11] W. C. Hung, K. L. E. Law, and A. Leon-Garcia, “A Dynamic Multi-Channel MAC for Ad Hoc LAN,” In Proc. 21st Biennial Symp. Comm, June 2002, pp. 31-35.
[12] A. Tzamaloukas and J.J. Garcia-Luna-Aceves, “Channel-Hopping Multiple Access,” In Proc. IEEE Int’l Conf. Comm. ICC ’00, June 2000, pp. 415-419.
[13] A. Tzamaloukas and J. Garcia-Luna-Aceves, “Channel-Hopping Multiple Access with Packet Trains for Ad Hoc Networks,” In Proc. IEEE Device Multimedia Comm.
MoMuC ’00, Oct. 2000.
[14] J. Chen, S. Sheu, and C. Yang, “A New Multichannel Access Protocol for IEEE 802.11 Ad Hoc Wireless LANs,” In Proc. 14th IEEE Int’l Symp. Personal, Indoor and Mobile Radio Comm. PIMRC ’03, vol. 3, Sept. 2003, pp. 2291-2296.
[15] P. Bahl, R. Chandra, and J. Dunagan, “SSCH: Slotted Seeded Channel Hopping for Capacity Improvement in IEEE 802.11 Ad Hoc Wireless Networks,” In Proc. ACM MobiCom’04, Sept. 2004.
[16] S. Misra, M. Reisslein and X. Guoliang, “A survey of multimedia streaming in wireless sensor networks,” IEEE Communications Surveys & Tutorials, Vol. 10, 2008, pp. 18-39.