ES-MAC: Energy Efficient Sensor-MAC Protocol for Wireless Sensor Networks Shama Siddiqui Faculty of Computer Science Institute of Business Administration Karachi, Pakistan
[email protected]
Abstract—Energy management is one of the main r e s e a r c h issues associated with Wireless Sensor network. An optimum means of preserving energy must always be deployed. Energy management in networks is also considered as one of the most crucial topics in the development of green networks. In this paper, we propose a modified duty-cycle based MAC protocol named as ES-MAC (Energy Efficient Sensor MAC). It involves energy efficient mechanism for wireless sensor motes by making them sleep when idle and by reducing the number of transmitted packets using Selective Data Transmission (SDT). Sleep intervals are improved by using the concept of Dynamic Data Cycle ( DDC). The proposed protocol is shown to be efficient by test bed implementation, which is our major contribution. We compared the results of ES-MAC with two control cases. TinyOS and Crossbow motes are used for implementing the nested C code. Keywords- motes; energy efficient; synchronization; Dynamic Data-Cycle;
I.
INTRODUCTION
Wireless Sensor Networks (WSNs) play a vital role in a variety of applications such as environmental monitoring (by sensing parameters such as temperature, light, pressure, humidity, etc), industrial & building automations, mobile asset management, and physical security [1]. As more application areas are being developed, effective energy management of these networks becomes essential, especially toward the development of green networks. Such networks consist of a large number of tiny nodes capable of sensing, computing and communicating [1-2], and hence there is high amounts of energy consumption for operation of such networks. The nodes are typically powered by batteries with a limited lifetime. Due to risk of network dis-connectivity or data loss, it is not desirable to replace the batteries frequently [3]. This is due to the possibility that during the replacement process, some data may get lost; therefore, for avoiding data loss and achieving long battery life, some methodology must be used [4]. The proposed protocol ES-MAC aims at optimizing energy efficiency of the practical applications of Wireless sensor nodes, where it is not feasible to change the batteries for the period of years. Since the protocol utilized two schemes, Dynamic Data Cycle and Selective Data Transmission, it would be helpful for the sensor nodes deployed for the applications of monitoring or event detection.
Sayeed Ghani Faculty of Computer Science Institute of Business Administration Karachi, Pakistan
[email protected]
The rest of the paper is organized as follows. In Section II, we present the related literature review. Section III explains the proposed protocol ES-MAC. Section III A discusses the synchronization scheme that the protocol utilizes. In Section III B, the ES-MAC protocol is explained with the help of timing diagrams. Section III C and D present the evaluation of ESMAC, with the help of two control cases. In section IV, we present the experimental results, and finally we present our conclusions and possible future extensions in Section IV. II.
LITERATURE REVIEW
During the last decade, Considerable amount of research have been done for improving energy efficiency of wireless sensor networks. The main approaches for efficient energy management in sensor networks with energy-hungry sensors are surveyed in [1]. The paper describes the fact that in many practical applications, the power consumption of the sensing activity may be comparable or even greater than that of the transmitting radio. Therefore effective energy management strategies should include policies for saving energy in the data acquisition and processing stage as well as in the transmission stage. For accomplishing the above, the two main approaches duty-cycle and adaptive sensing are proposed, as described below. These schemes reduce the number of transmitted packets, which should result in saving energy. Duty cycle based protocol was first introduced in [5] to reduce energy consumption during idle listening. A novel protocol known as S-MAC was proposed, in which each sensor node follows a periodic wake-up/sleep schedule in synchronization with each other. Each such cycle is divided into a Data period and a Sleep period. During the data period, after synchronization stage, data packets are sent to the single hop neighbors, and nodes go to sleep simultaneously for the prespecified sleep period. The main disadvantage of this protocol was the increased end-to-end latency, as the packet only reaches to the single hop destination after a single duty cycle. T-MAC, a contention based protocol introduced the concept of adaptive duty cycle to deal with the idle listening problem [6]. In T-MAC, the Data period is classified as active or nonactive. When non-active, the nodes go to sleep state, even before the completion of data cycle. As the wake up duration is reduced for T-MAC, on comparison with S- MAC, it is shown to be more energy efficient.
Protocols such as R-MAC [7], DW-MAC [8], PRMAC [9], BulkMAC [10] are based on the concept of using routing layer information to create multi hop and multi packet flows. These protocols send a PION packet during the data cycle consisting of information such as source address, destination address, previous hop, next hop, total number of hops, and number of packets each node has to send, so that any particular node can add its own packets while forwarding the nodes from the previous node. Actual data is sent in the sleep cycle. The nodes calculate their sleep duration depending upon the information given in the PION packet. These protocols significantly reduce end-to-end latency and result in a remarkable improvement of energy efficiency. Scheduled wake-up methods based on cross layer information are proposed in [11]. Multi parent schemes which assign different parents with different wake-up schedules are proved to be energy efficient and good at managing latency. A comparative study and comprehensive summary of the available MAC protocols (S-MAC, T-MAC, TA-MAC, PMAC, O-MAC, etc) for wireless sensor networks is presented in [12]. III.
PROPOSED PROTOCOL: ES-MAC
In this paper, we propose an energy saving mechanism for wireless sensor networks. A MAC protocol called ES-MAC is proposed, whose objective is to reduce the number of packets sent, and also to allow the mote to sleep for the time when it is idle even in the data cycle. This protocol is essentially a modification of S-MAC [2] and T-MAC [3], yet novel because it utilizes the concept of Dynamic Data Cycle (DDC) and Selective Data Transmission (SDT) combined. Under the normal practice, the sender node sends data after pre-specified intervals of time without checking the application data values. However this makes the system inefficient in terms of energy usage, because same data values are being sent repeatedly. It is evident that data values that are not changing may need to be sent again and again, we propose that even in the wake-up state, nodes should check the data and must not broadcast until they have some significantly changed data values. By implementing this protocol significant amount of energy can be saved, because depending on the application, the number of packets transmitted may be reduced drastically. It must be noted that, this research mainly aims at the introduction of DDC and SDT in an elementary transmission protocol ES-MAC. The protocol does not consider the advanced issues, such as node discovery, multiple schedules of the nodes or network allocation vectors. Our major contribution is to implement the proposed modifications on the test bed using IRIS motes. Control cases are usually used to evaluate the efficiency of proposed models. For evaluating the efficiency of ES-MAC protocol, we compared it with two control cases; As ES-MAC makes two changes to the legacy S-MAC, i.e; a Dynamic Data Cycle and secondly, Selective Data Transmission. We evaluated the effects of both parameters. Hence we have 3 scenarios as follows:
i.
ES-MAC: Uses Dynamic Data Cycle (DDC) and Selective Data Transmission (SDT), discussed in Section III B.
ii.
ES-MAC (Control Case 1) - ES-MAC without DDC and SDT, discussed in Section III C.
iii.
ES-MAC (Control Case 2) - ES-MAC with DDC but without SDT, discussed in Section III C.
As the nodes will be periodically sleeping, it is essential for the motes to be awakened simultaneously, to avoid packet loss. Therefore, some scheme of synchronization must be utilized. Prior to the discussion of the three scenarios, the proposed synchronization scheme is explained briefly, in Section III A. A. Synchronization A simplistic scheme is used for synchronization that also results in reduction of the packet overhead. In the other duty cycle based protocols, additional overhead is involved due to the transmission of synchronization packets. ES-MAC does not use any synchronization packet; rather it has a synchronization period when the nodes initially wake up. Both, the sender and receiver will initially remain in wake up state until sender transmits the first packet and receiver acknowledges it. This duration is termed as synchronization cycle. After which, they will simultaneously go to sleep for the sleep duration (Ts) as shown in Figure 1a. B. Working Mechanism of ES-MAC Once synchronization is completed, the motes alternate between two cycles referred to as data-cycle and sleep-cycle. The data-cycle is of duration TD, and sleep-cycle of duration Ts. The sender mote will be switching between the states of “sleep” and “wake-up” depending upon duty-cycle scheme. In our system, Duty cycling consists of waking up the sensor system only for the time needed to acquire a new set of samples and going to sleep immediately after broadcasting and receiving the acknowledgements as shown in figures 1a and b. During the wake-up time only sufficiently changed data values will be broadcasted, hence saving the energy of processing as well as transmission. We refer to this type of data transmission as Selective Data Transmission (SDT), and scheduling for sleep and wake-up cycle as Dynamic Data Cycle (DDC), because once a new data value is sent, the node will immediately go to sleep without waiting for the pre specified wake up duration to complete. While using this energy-efficient methodology, it would be necessary for the sender node to receive the acknowledgement after which it may go to sleep. In case of acknowledgement not received, the data should be transmitted repeatedly, as shown in the timing Diagram for ES-MAC (figure 1a & b).
As shown in figure 1b, we consider two nodes S and R, representing sender and receiver respectively. Here S is responsible for sensing, detecting the data change, and broadcasting the data. “D” represents a transmitted packet and “A” represents an acknowledgement packet. The description of the variables is as follows:
cycle of the same duration Ts begins at the same instant and hence, the nodes become synchronized. Ideally, using this scheme, synchronization should remain continue to be maintained. Now, for the sake of illustration of the working of this protocol, Let us consider 3 Data/Sleep cycles as follows. During the first cycle, S is assumed to remain in wake-up state for a period of TD. For this duration, S has to sense for data change at every Tr period, and transmit. Consider, when S senses first time in the Data period, it detects the changed value and hence transmits.
Tr = Time for reading data TD = Data Cycle Ts= Sleep Cycle Trw= Remaining time in wake-up duration
(a)
(b) Figure 1(a) Duty Cycle for ES-MAC, (b) Timing Diagram for ES-MAC
When a node initially wakes up, synchronization period starts before the beginning of data cycle. During this period, assuming that node S broadcasts the packet destined for R, but R was not awake or for some other reason, S failed to receive acknowledgement. Due to which, it will remain in wake-up state as long as it does not receive an acknowledgement, and will continue to sense and send data at each Tr sec. For this particular case, S will keep sensing for changed value, but even if the value remains unchanged, it will be sent. As soon as R sends an acknowledgement, both S and R go to sleep for Ts sec. As transmission and reception of an acknowledgement packet takes place almost at the same instant, so their sleeping
As soon as R sends the acknowledgement, both S and R will go to sleep. In this case, since the nodes go to sleep well before the completion of pre specified interval TD, therefore sleep duration is longer than Ts in order to keep the nodes synchronized. Assuming for this particular cycle, the duration of Trw sec was still remaining in the completion of TD, hence the sleeping time would be Sleeping time = Ts + Trw
(1)
Equation (1) represents the concept of DDC, on which ES-MAC mainly relies for becoming energy efficient.
For the second cycle, S sent the packet but now we assume that acknowledgement was not received. The packet will be retransmitted, and when acknowledgement is received, we assume it is exactly the end of TD, and so both nodes will go to sleep for Ts. However if in any particular cycle, R fails to acknowledge for the entire TD, the situation may result in data loss. To accommodate such condition, on next wake up, S will first check for data change, if it has changed it will be sent. If data is not changed, then S will check for the status of previous acknowledgement. In case of no acknowledgement, it will resend the same packet until acknowledgment reception. For the third cycle, it is assumed that upon wake up, no data change was detected, due to which there was nothing to send. S continues to read after Tr, for the complete wake up duration, at the end of which, nodes will go to sleep. Working of ES-MAC sender and receive is summarized in figures 2a and b respectively. It shows how the protocol continues to work after the nodes get synchronized. C. Control Case 1 This case is used to evaluate the performance of both the parameters proposed by ES-MAC. In this case, neither of the features: DDC and SDT are used. Hence, the nodes will always follow the pre-specified intervals of wake up and sleep intervals, and will send every data value without monitoring for change. D. Control Case 2 This case is used to evaluate the performance of the parameter of SDT proposed by ES-MAC. In this case, all of the features of ES-MAC are kept same, except that S will send all data without SDT. Nodes will follow the DDC as discussed in Section III B. Timing Diagram for this case is shown in figure 4. Nodes get synchronized in the same way as described in Section III A. For the first cycle, acknowledgement was not received for the first data packet, but S will keep on sending the new packets which may be new or the previous one. Upon receiving Acknowledgement, both nodes sleep, following DDC. In cycles 2 and 4, acknowledgements were received for the first transmission attempts, and hence the nodes sleep immediately. While, the third cycle is a replication of first cycle; as two packets are sent before the nodes sleep.
Wake up for TD sec
Read & display data after Tr sec
No
Is data changed?
Wake-up Duration Complete?
No
Yes Yes
Broadcast with Receiver’s Address
Go to Sleep for Ts sec
Yes No
Is Ack received? Yes
Go to Sleep for Ts sec + Remaining time in Wakeup (Trw)
Sleep Duration Complete? Yes
Figure 2(a). ES-MAC Sender
Wake up and Start Listening for Tr sec.
Is data Received?
No
Yes
Wake-up Duration Complete?
No
Yes
Send Ack, Display Data Go to Sleep for Ts sec Go to Sleep for Ts sec + Remaining time in Wakeup
Figure 2(b). ES-MAC Receiver
No
Figure 3. Timing Diagram for Control Case 1
Figure 4. Timing Diagram for Control Case 2
IV.
EXPERIMENTAL RESULTS AND DISCUSSION
The results are obtained using TinyOS environment and Crossbow motes, along with Sensor board 1 [13]. The motes had light and temperature sensors. For this paper, only light sensor data was used. The aim of the experiment is to evaluate the efficiency of the proposed protocol. The number of packets transmitted is observed for the 3 scenarios discussed in Section III. We assumed the following numerical values for experiment. TD = 5 sec, Tr = 2 sec, and Ts = 10 sec. For simplicity, we used a single sender in combination with a single receiver. For all three scenarios, we monitored the number of transmitted and acknowledged packets in two situations; for low and high variations of light sensor data. For ES-MAC, we further considered two scenarios; the first when data has to change by more than 0.5% for transmission to take place and the second when data has to change by more than 1% for transmission to take place.
1
Details of Crossbow motes are as follows: o MIB520 Gateway o IRIS-XM2110 o MDA-100 CB Sensor board, with light and temperature sensors
For low or normal variation, a typical office or home environment is monitored and the experiment is performed for duration of 60min, while for forced or high variations, light data is forcibly varied and experiment is performed for duration of 20 minutes. To make the results comparable, number of Packets transmitted Per Minute (PPM) is observed as shown in Table1. Number of packets is found to be lowest for ES-MAC with both conditions, i.e.; as soon as S will detect any new data (depending upon the level of change required), it will immediately go to sleep after transmitting, thus, increasing the duration of sleep interval. Secondly, S will not broadcast any packet until it finds some change. When ES-MAC is tested for low variation of data (0.5% change), the PPM found to be 3.4, which is the lowest of all cases. The value is, as expected because the protocol sends the packets which satisfy both necessary conditions. For different level of percentage changes, the number of packets further decreases, as for 1% change level, the PPM reduces to 2.99. For high variations, the number of packets increases, as theoretically expected. The difference between the PPM at different percentages proves the efficiency of protocol in its SDT feature.
TABLE I. COMPARISON OF NUMBER OF PACKETS TRANSMITTED AND ACKNOWLEDGED FOR ES-MAC & ES-MAC CONTROL CASES 1 & 2 Level of Variation Low (Normal) Duration = 60min
Category
High (Forced) Duration = 20min
Data
Ack
Total
Total/min
Data
Ack
Total
Total/min
ES-MAC (0.5% change)
102
102
204
3.4
76
76
152
7.6
ES-MAC (1% change)
88
88
176
2.99
51
51
102
5.1
ES-MAC (Control1)
491
491
982
16.4
164
164
328
16.4
ES-MAC (Control2)
243
243
486
8.1
81
81
162
8.1
For the control case 1, when S sends all data for the prespecified wake up duration, the number of packets transmitted per minute is found to be the greatest (16.4), as expected. As per our numerical assumptions, in each cycle of 15 sec (TD + Ts), 2 packets are transmitted and 2 acknowledged. (Data read and sent at every 2 sec, while TD itself is 5sec).We noted number of packets transmitted by both S and R. The reading of PPM as 16.4 corresponds to about 4 packets over 14.6 sec, which is approximately as expected. For control case 2, S sends the packet without SDT but it follows the DDC. In each cycle, as soon as S first sends the packet after 2 sec, and receives an acknowledgement, it immediately goes to sleep, reducing the number of transmitted packets per cycle to 1. Thus implementing DCC causes the reduction of 50%, as compared with control case. As both of the control cases don’t monitor the data change, so there is no difference in number of transmitted packets for low and high data variations. V.
CONCLUSION AND FUTURE WORK
The proposed protocol ES-MAC is shown to be efficient in energy usage as it drastically reduces the number of transmitted packets. The protocol utilizes the concepts of SDT and DDC, which is our main contribution. The number of transmitted packets depends on these two parameters. The results of ES-MAC show 80% reduction in number of transmitted packets, when the protocol deploys DDC and SDT both. If the protocol uses only DDC, the number of packets transmitted reduces by 50%. This proves the energy efficiency of the protocol for transmission. Thus, the protocol is of extreme importance for the practical deployment, where it is essential to conserve energy in order to prolong the battery life. For ease of observation and explanation of the model, we made certain major assumptions; For example, we assumed the value of Ts to be exactly double of TD. However to make the system energy efficient and enhancing the battery life up to years practically, Ts has to be kept much larger than There may be many dimensions for extending this work. Some of the examples include; considering back-off mechanisms while implementing ES-MAC, implementing multi hop scenarios, considering separate schedule tables for multiple nodes environment, Generating sensor based interrupts for making the nodes wake up, and taking drastic data changes into account.
REFERENCES Allipi C, Anastasi G., Di Francesco M., Roveri M., “Energy Management in Wireless Sesnor Networks with Energy Hungry Sensors”, Instrumentation & Measurement Magazine, IEEE, vol. 2, pp. 16-23, April 2009. [2] S. Mohanty, “Energy Efficient Routing Algorithms for Wireless Sensor Networks and Performance Evaluation of Quality of Service for IEEE 802.15.4 N e t w o r k s ”, N a t i o n a l I n s t i t u t e o f Technology, Rourkela, Orissa, India,Masters thesis, 2010. [3] I. Dietrich, F. Dressler, “On the Life Time of Wireless Sensor Networks”, ACM Transactions on Sensor Networks, Vol. 5, No. 1, January 2009, pp. 1-38. [4] B. Yahya and J. Ben-Othman, “Towards A Classification of Energy Aware MAC Protocols for Wireless Sensor Netwroks”, Wireless Communications and Mobile Computing, 2009. [5] W. Ye, W. Ye, J. Heidemann, and D. Estrin, “An Energy-Efficient MAC Protocol for Wireless Sensor Networks,” in Proceedings of IEEE INFOCOM 2002, Jun. 2002, pp. 1567–1576. [6] T. van Dam and K. Langendoen, “An Adaptive Energy-Efficient MAC Protocol for Wireless Sensor Networks”, in Proeedings of. First International Conference on Embedded Networked Sensor Systems (SenSys 2003), Nov. 2003, pp. 171-180. [7] S. Du, A. Saha, and D. Johnson, “Rmac: A Routing-Enhanced DutyCycle Mac Protocol for Wireless Sensor Networks,” in INFOCOM 2007. 26th IEEE International Conference on Computer Communications, pp.1478– 1486, May 2007. [8] Y.Sun, S. Du, O. Gurewitz and D. B. Johnsun, “DW-MAC: A Low Latency, Energy Efficient Demand-Wake Up MAC Protocol for Wireless Sensor Networks”, Proceedings of 9th ACM International symposium on Mobile adhoc networking and computing Hong Kong, Hong kong, China: ACM, 2008. [9] T. Canli and A. Khokhar, “Prmac: Pipelined Routing Enhanced MAC Protocol for Wireless Sensor Networks,” In 2009. ICC ’09. IEEE International Conference on Communications, pp. 1–5, June 2009. [10] T. Canli, M. Hafeida, and A. Khokhar, “BulkMAC: A Cross-Layer Based Mac Protocol for Wireless Sensor Networks”, in Procedings of the 6th International wireless communications and Mobile Computing Conference, IWCMC’10, pp. 442-446, 2010. [11] A. Keshavarzian, H. Lee, and L. Venkatraman, “Wakeup Scheduling in wireless Sensor Networks”, in Proceedings of 7th ACM International Symposium on Mobile Ad hoc Networking and Computing (MobiHoc 2006), May 2006, pp. 322-333 [12] M. I. Shukur, L. S. Chyan and V. V. Yap “Wireless Sensor Networks:Delay Guarentee and Energy Efficient MAC Protocols”, Proceedings of World Academy of Science, Engineering and Technology, WCSET 2009, 25-27 Feb. 2009,Penang, Malaysia. [13] The Memsic Solution, http://www.memsic.com/products/wirelesssensor-networks/wireless-modules.html [1]
