An Energy Efficient and Delay Sensitive Centralized MAC Protocol for Wireless Sensor Networks Celal CEKEN Kocaeli University, Technical Education Faculty Electronics and Computer Education Department 41380 Kocaeli, Turkey Telephone: +90 262 303 22 40 Fax: 0 262 3058010 [email protected]Abstract. Energy consumption is one of the most crucial design issues in wireless sensor networks since prolonging the network lifetime depends on the efficient management of sensing node energy resource. In this research study, a new TDMA based MAC protocol, which is not only energy aware but also delay sensitive, is introduced for wireless sensor networks. In the proposed MAC, to achieve energy conservation, sensing nodes employing the proposed MAC sleeps periodically to reduce duty cycle and minimize idle listening. In addition, to provide lower message delay, time critical sensing nodes request extra time slots form the central node. Unlike common wireless sensor network models with a multi-hop topology, the proposed WSN architecture has a centralized structure especially for energy efficiency and fulfillment of the delay requirement of time critical networking applications. The proposed MAC
36
Embed
An Energy Efficient and Delay Sensitive Centralized MAC Protocol for Wireless Sensor Networks Just Before Submit Ti On
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
An Energy Efficient and Delay Sensitive Centralized MAC Protocol for Wireless Sensor Networks
Celal CEKEN
Kocaeli University, Technical Education Faculty Electronics and Computer Education Department
Fig. 5. (a) Connection request packet, (b) Connection replay packet, (c) Data packet, (d) Extra slot request packet, (e) Extra slot replay packet.
When the CN gets the connection request packet, if the resources are sufficient, it allo-
cates a time slot for the request and sends the slot number to the related SN using the
connection replay packet (Fig. 5b). After an SN gets its time slot(s),which means con-
nection has been established, the data sensed is conveyed with the data packet illus-
trated in Fig. 5c in its own time slot(s). A data packet comprises 48 bytes, consisting of
a 1–byte header (SourceID), and a 47–byte information field for sensed data. If an SN
needs more bandwidth for delay sensitive traffics it requests extra time slot again from
the CN using extra slot request packet (Fig. 5d). If there are adequate number of empty
slots, CN allocates one more time slot and sends the slot number to the related SN using
the extra slot replay packet (Fig. 5e). A 2-byte error correction field (CRC) which is
used for detection and correction of the possible packet transmission errors is also
added to all data packets. The process model of the proposed WSN MAC module em-
ployed in SN is illustrated in Fig. 6.
Fig. 6. The SN MAC layer process model.
The process starts with the big arrow, pointing the init state. This state performs a delay
until the other processes in the simulation are initialized and loads the control variables.
Then the process enters the idle state and waits here until a specific interrupt arrives.
The conReq state machine creates connection request packet informing connection es-
tablishment and sends it to the CN. The reqResp state machine obtains the number of
time slot assigned by the BS. The fromSrc state machine gets the data sensed from the
upper layer, segments it to the data packets and inserts them into the queue. The data
packets received from the upper layer are sent to destination in the time slot(s) dedi-
cated to the SN in toTX state machine. The sleep state machine turns off the radio
functions for a specific time interval to conserve energy. The extSlotReq state machine
creates extra slot request packet to inform extra bandwidth requirement for delay sensi-
tive traffics. The fromRx state machine delivers any arrived packets destined to the SN
to MAC layer. All SN MAC layer process model functions are outlined in Fig. 7.
State Machines
init
idle
conReq
fromSrc
reqRep
fromRx
toTx
Initialize
Wait for an interrupt.
Connection request
Sensed dataarrival
Control packet from CN
Slot && DataENQ
Create connection request packet and send it to the CN in f irst empty slot.
If the connection has been established, then put the data sensed to the packet.
SNs slot time
Send the packet.
Set up an interrupt for the SN next slot time.
Destroy received packet.
Packet for this SN
Get the SNs slot from the packet and destroy
extraSlotReqExtra slot
request
Create extra slot request packet and send it to the CN in f irst empty slot.
Y
N
Y
Y Y
Y
Y
Y
N
N
N
N
N
N
Functions
sleepSleep mode
Turn off the radio for 50 sec.
Y
NSleep mode
off
If the connection has been established, then put the data sensed into the buffer.
YN
Sensed dataarrival
Wait for an interrupt.
Y
N
Fig. 7. The SN MAC layer process model algorithm.
4.2 Central Node MAC Model
The CN gathers all the data sensed from the environment by the SNs in the cluster and
coordinates how the SNs will access the wireless medium fairly. The CN functions of
the proposed MAC protocol include three main processes. These are namely, assigning
time slot for any SN, forwarding any arrived data packets to upper layer and allocating
extra time slots for delay sensitive data traffics using the ST scheduling algorithm. Fig.
8 shows the proposed CN MAC model realized using OPNET Modeler.
Fig. 8. The CN MAC layer process model.
Like in the former process model, this process also starts with the init state, then enters
the idle state and waits here until a specific interrupt arrives. The fromRx state machine
delivers arriving packets to the next state machine according to the packet formats. The
bwReq state machine handles connection and extra time slot requests and also executes
a fair scheduling algorithm that manages the ST which stores the SNs slot usage infor-
mation. The data state machine delivers the sensed information to the upper layers to
execute a specific task. The CN MAC layer process model algorithm is outlined in Fig.
9.
State Machine
init
idle
bwRequest
fromRx
data
Initialize
Wait for interrupt
Packet from SNs
Destroy received packet.
N
N
Sensed data packet
Get the packet and deliver it to upper layer.
Connectionrequest
Extra slot request
Allocate a slot for the related SN. Create a connection response packet and send it to the SN.
ES
N
ES :Is there any empty slot (i.e. in ST dedicated=0 )LPS :Is there any low priority slot (i.e. in ST priority=0 )
Y
Allocate a slot for the related SN. Create a connection response packet and send it to the SN.
LPS
N
Y
ES
N
Y
Allocate a slot for the related SN. Create a extra slot response packet and send it to the SN.
Y
Functions
Y
Y
Y
N
N
Fig. 9. The CN MAC layer process model algorithm.
5. Computer Simulation of WSN
In the example scenario shown in Fig. 10, to generate sensed data traffics there are sev-
eral SNs which are implemented using OPNET Modeler and employing the proposed
MAC protocol explained in the previous section. The sensed data traffic introduced to
the network by any SN is destined to the CN for executing a specific task. Some of
these nodes considered as generating delay sensitive application traffics while the others
generating non time critical data traffics. Diameter of the cluster which constructs the
network topology is chosen 100 meters.
Fig. 10. Example WATM scenario.
Another WSN model analogous to the one above except that IEEE 802.11 DCF MAC
protocol is utilized instead of the proposed MAC is also simulated using OPNET Mod-
eler. Working conditions of both network models are chosen to be same for consistent
performance comparisons.
5.1 Simulation Results and Discussion
Simulation results of the both WSN models described above are presented under vary-
ing network load conditions followed by performance comparisons and analysis. In the
simulation environment a free space channel propagation model that supports to predict
received signal strength when the transmitter and receiver have a clear, unobstructed
line-of-sight path between them is utilized.
In the proposed MAC, an uplink frame consists of 125 time slots each has 1 ms length
and contains 2 data packets. The simulation parameters used are given in Table 1. The
simulation was run for 3600 second.
Table 1: Simulation parameters
Parameter Value
Message Size 20 packets x 50* Bytes
Interarrival Time 1*-10* sec
Data Rate 1 Mb/s
Frequency Band Uplink=3 GHz and Downlink=4 GHz
Transmitter Power CS= 10 mW and SNs=10 mW
Modulation Schema BPSK
Number of SNs 100
Area size 100 m x 100 m
Channel Model Free Space Propagation Model (LoS)
*Generated using Exponential Distribution Function Exp(Mean).
In most of the previous research related to WSNs, the major goal is to minimize the
power consumption of SNs. However, the focus of this work is not only improving the
power conservation performance but also providing a better average packet transfer de-
lay for especially time critical application traffics. The packet loss ratio metric is not
considered here since the buffers are assumed to have enough capacity so that no data
packet is lost due to buffer overflow. Moreover, it is also assumed that the CRC bits
added to the packets avoids the possible bit errors.
In the example scenario, while most of the SNs are deployed to generate non time criti-
cal sensed data, the SN1 is deployed to produce delay sensitive application traffic. All
non time critical SNs put themselves into sleep mode after 50 seconds of being idle and
stay this mode for next 50 seconds and this process repeats during the simulation run
time. Varying the message size of all SNs application traffics, power consumption and
average EED (end-to-end delay) results of the delay sensitive traffic transfer between
SN1 and CN, of non time critical traffic transfer between SN2 and CN have been col-
lected during the simulation run time for both WSN models.
In the proposed MAC, there are two parameters that effect the power consumption and
latency performance of the SNs. The first is the sleep mode, for non time critical SNs,
on which the power consumption ratio is considerably reduced while it results in in-
creasing end to end message delay. The second is the extra slot usage for delay sensitive
SNs, which provides lower latency performance but conversely results in higher power
consumption ratio due to the increasing channel utilization.
In Fig. 11, average EED results of the WSN models are presented as a function of the
interarrival time. To compare the results easily, Fig. 12 also shows the average EED re-
sults for the proposed MAC based WSN model, which are normalized with those of the
IEEE 802.11 DCF MAC based WSN counterparts. For heavy traffics (i.e. interarrival
time is up to 3 sec), in the best case, the delay sensitive application traffic (i.e. between
SN1 and CN) experiences approximately 15000 times lower (due to the burst slots em-
ployed in the proposed MAC), the non time critic application traffic (i.e. between SN2
and CN) experiences approximately 299 times lower average message delays in the pro-
posed MAC based WSN model compared to the same traffic carried with the IEEE
802.11 DCF MAC based WSN model. However, for the light traffics (i.e. interarrival
time is between 3 sec and 8 sec) the EED results of the IEEE 802.11 DCF MAC are
generally better than those of the both proposed MAC models. Moreover, it can be ob-
served from the figure that, for the proposed MAC, the longer duty cycle results in an
increase in message delay as expected.
0.01
0.1
1
10
100
1000
10000
1 2 3 4 5 6 7 8Interarrival Time (s)
Ave
rag
e E
nd
-to
-En
d D
elay
(s)
IEEE 802.11 DCF MAC Proposed MAC SN1-CN (delay sensitive) Proposed MAC SN2-CN (w ith sleep mode)
Fig. 11. Average EED results of the MAC protocols.
0.00001
0.0001
0.001
0.01
0.1
1
10
100
1000
1 2 3 4 5 6 7 8Interarrival Time (s)
Ave
rag
e E
nd
-to
-En
d D
elay
(s)
IEEE 802.11 DCF MAC Proposed MAC SN1-CN (delay sensitive) Proposed MAC SN2-CN (w ith sleep mode)
Fig. 12. Normalized average EED results of the MAC protocols.
In Fig. 13, measured average power consumption results of the WSN models are pre-
sented as a function of the interarrival time. Fig. 14 also shows the average power con-
sumption results for the proposed MAC based WSN model, which are normalized with
those of the IEEE 802.11 DCF MAC based WSN counterparts. As it can be seen from
the figure, power consumption results of the proposed MAC are better than those of the
IEEE 802.11 DCF MAC for all traffic conditions. In the best case, non time critical
SN2 employing the proposed MAC consumes 1.8 times lower energy than the one em-
ploying the IEEE 802.11 DCF MAC. For the proposed MAC model, non time critical
SN2 provides 1.1 - 1.7 times lower power consumption than SN1. This is not a surpris-
ing outcome since SN1 uses extra slot(s) to accomplish better latency performance and
this consequently results in increasing channel utilization, meaning additional power
consumption. Besides, SN2 puts itself into the sleep mode periodically and this pro-
vides a significant amount of reduction in power consumption ratio because the idle and
transmit modes consume rates are higher than the sleep mode consume rate.
4
5
6
7
8
9
10
11
1 2 3 4 5 6 7 8Interarrival Time (ms)
Ave
rag
e P
ow
er C
on
sup
tio
n (
mW
)
IEEE 802.11 DCF MAC Proposed MAC SN1-CN (delay sensitive) Proposed MAC SN2-CN (w ith sleep mode)
Fig. 13. Average power consumption results of the MAC protocols.
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1 2 3 4 5 6 7 8Interarrival Time (ms)
Ave
rag
e P
ow
er C
on
sup
tio
n (
mW
)
IEEE 802.11 DCF MAC Proposed MAC SN1-CN (delay sensitive) Proposed MAC SN2-CN (w ith sleep mode)
Fig. 14. Normalized average power consumption results of the MAC protocols.
For the network model employing the proposed MAC, when an SN enters a sleep
mode, its transmission, receiving and idle operations are halted. Thus, the sensed data is
accumulated in the buffer. In Fig. 15, queuing statuses of SN1 and SN2 are shown. As
it can be seen, size of the data in the SN2 queue is more than that of SN1queue during
the simulation run time as a consequence of the sleep mode operation and this result in
increasing message transfer delay.
100
1000
10000
100000
0 10 20 30 40 50 60
Simulatin Time (minutes)
Qu
eue
Siz
e (b
ytes
)
Proposed MAC SN1 (delay sensitive) Proposed MAC SN2 (w ith sleep mode)
Fig. 15. Queuing statuses of SN1 and SN2 with the proposed MAC.
5. Conclusions
Many ongoing researches on WSN subject focus only on the energy efficiency. In this
study a new energy aware and delay sensitive MAC protocol for WSNs has been pro-
posed and simulated using OPNET Modeler software. The simulation results have also
been compared with those of the IEEE 802.11 DCF MAC protocol. According to the
performance results obtained, with the scheduling algorithms developed for the pro-
posed MAC protocol, not only have lower energy consumption ratios been fulfilled but
also lower end to end message delay results have been achieved for especially delay
sensitive data traffics. For the proposed MAC, in the best case, 1.8 times lower energy
consumption results and 15000 times lower latency performance have been obtained
when compared with those of IEEE 802.11 DCF MAC protocol.
Acknowledgement
The author would like to thank Assoc.Prof.Dr. Ismail Erturk for his invaluable contri-
butions to this study.
References
[1] Editorial, Wireless sensor networks designs, Computer Communications (2005) 1-4.[2] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, A survey on sensor networks, IEEE
Communications Magazine 40.8 (2002) 102-114.[3] W. Ye, J. Heidemann and D. Estrin, An Energy-Efficient MAC Protocol for Wireless Sensor Net-
works, Twenty-First Annual Joint Conference of the IEEE Computer and Communications Soci-eties (INFOCOM) 3 (2002) 1567-1576.
[4] P.C. Nar, E.Cayirci, PCSMAC: A Power Controlled Sensor-MAC Protocol for Wireless Sensor Networks , IEEE Conf. (2005) 81-92.
[5] Vaduvur Bharghavan et al., MACAW, ACM SIGCOMM Computer Communication Review, Proc. Conf. on Communications Architectures, Protocols and Applications 24.4 (1994) 212–225.
[6] T. van Dam and K. Langendoen, An adaptive energy-efficient MAC protocol for wireless sensor networks, In The Firsr ACM Conference on Embedded Networked Sensor Sysiems (2003).
[7] G. Lu, B. Krishnamachari, C.S. Raghavendra, An adaptive energyefficient and low-latency MAC for data gathering in wireless sensor networks, Proceedings of 18th International Parallel and Dis -tributed Processing Symposium (2004) 224.
[8] C.C. Enz, A. El-Hoiydi, J-D. Decotignie, V. Peiris, WiseNET: An ultralow-power Wireless Net-work Solution, IEEE Computer 37.8 (2004).
[9] I. Demirkol, C. Ersoy, F. Alagöz, MAC Protocols for Wireless Sensor Networks: A Survey, IEEE Communications Magazine, (2006) 115-121.
[10] M. Stemm, R. H. Katz, Measuring and reducing energy consumption of network interfaces in hand-held devices, IEICE Transactions on Communications, E80-B 8 (1997) 1125–1131.
Celal CEKEN received the M.Sc. and Ph.D. degrees from Kocaeli Univer-sity, Turkey in 2001 and 2004, respectively. His active research interests include wireless communications, broadband networks, ATM networks, high speed communication protocols, and Wireless Sensor Networks.