Extending Beacon-Enabled IEEE 802.15.4 to achieve Efficient Energy Savings: Simulation-Based Performance Analysis Mounib Khanafer, Mouhcine Guennoun, Hussein T. Mouftah School of Information Technology and Engineering University of Ottawa 800 King Edward Ave., Ottawa, ON, Canada
[email protected],
[email protected],
[email protected] Abstract—Wireless Sensor Networks (WSNs) operate in hostile environments under resource-constrained conditions. Power conservation is a primary factor that drives the design of these networks. Therefore, WSNs should not utilize complex algorithms that are power-hungry. The IEEE 802.15.4 standard is the appropriate suite of specifications that conforms to the distinguished characteristics of WSNs. This standard is suited for low data rate, low power, and low radio transmission ranges that are typical in WSNs. In this paper, we propose an extension to 802.15.4 that not only achieves efficient power savings, but also improves the reliability and the channel utilization in WSNs. In essence, we force each node that has just finished a successful packet transmission to sleep for a tunable period of time before contending for sending the next packet. We show through simulations that this behavior not only prolongs the lifetime of the WSN, but also achieves, compared to the original 802.15.4 standard, higher levels of channel utilization, better reliability, while preserving fairness among the nodes in the network. Index Terms— Wireless Sensor Networks, Beacon-Enabled IEEE 802.15.4, Channel Utilization, Power Conservation, Reliability.
I.
INTRODUCTION
The IEEE 802.15.4 (or 802.15.4 for simplicity) standard is the appropriate suite of specifications that conforms to the distinguished characteristics of Wireless Sensor Networks (WSNs). This standard is suited for low data rate, low power, and low radio transmission ranges [1] that are typical in WSNs. WSNs are deployed in hostile environments and it is impractical to recharge or replace the nodes that have depleted their power resources. If power conservation precautions have not been planned properly, the lifetime of the WSN is reduced substantially. Therefore, the conservation of power resources is a primary requirement that takes the highest priority whenever an algorithm is designed for WSNs. One common approach followed by sensor nodes to conserve power is to sleep while not involved in any communication activity. In this paper, we propose an extension to 802.15.4 that can achieve better power savings for sensor nodes. Our proposed extension is simulated in order to study its performance compared to the original 802.15.4 standard. The rest of the paper is organized as
follows. In Section II, we describe our proposed extension. In Section III, we review research contributions that worked on modifying the 802.15.4 standard to enhance its performance. Section IV describes the simulations we conducted to analyze the performance of the extended 802.15.4. Finally, Section V concludes the paper and provides future research directions. II.
EXTENDING THE IEEE 802.15.4 STANDARD
Our main objective is to find new opportunities for saving more power for sensor nodes, such that the lifetime of the overall WSN is prolonged. As we achieve that target, careful attention should be given to other parameters like channel utilization and reliability. These parameters are crucial in the assessment of the networks’ performance and any proposed extensions should not degrade their values. We mentioned earlier that nodes tend to sleep, whenever possible, to save their power resources. Therefore, we introduce a new sleep state, named, the standby (SB) state, which each node should enter after each successful transmission of a packet. In other words, as the nodes contend for medium access, once a node captures the medium and finishes transmitting its packet, the node should sleep for some duration of time, referred to as LSB (which can be tuned according to the network’s size as we discuss later). This behavior can effectively reduce the number of contending nodes and thus reduce the level of collisions (which is a power waster as nodes need to retransmit the lost packets). The benefit of this new SB state can be also sensed if we notice that many nodes can be in this state as the same time. This results in significant reduction of packet collisions. III.
RELATED WORK
Many proposals have been reported to enhance 802.15.4 in terms of different performance parameters, like throughput, packet delivery delay, and energy consumption. In [2], the authors propose enhancements for the nonbeaconed 802.15.4. Their enhancements target energyefficiency as well as reliable delivery of critical traffic. Basically, during network initialization, the duty cycle of the nodes is divided into active and inactive periods. During the active periods, nodes contend to access the medium according to the unslotted CSMA scheme. However, nodes
This paper has been accepted for publication in the proceedings of WSN Workshop that had been held in conjunction with NTMS 2011. 978-1-4244-8704-2/11/$26.00 ©2011 IEEE
sleep during the inactive periods. Nodes remain in the active mode while delivering critical data traffics. This behavior achieves higher reliability. However, inactive nodes allow nodes to conserve more power than the case is in the original non-beaconed 802.15.4, in which nodes are always awake and contending to access the medium. While this proposal outperforms the standard 802.15.4 as per the aforementioned parameters, it requires the employment of a synchronization scheme in the non-beaconed mode of 802.15.4. The overhead associated with that is not addressed in [2]. In [3], the authors propose DEEP, a MAC protocol for beacon-enabled 802.15.4 that optimizes the distribution of the guaranteed time slots (GTSs) such that better energy conservation is achieved. In particular, instead of having a GTS descriptor (that defines the node’s address as well as the GTS slot and direct) available in all subsequent beacons till the node requests its de-allocation, DEEP removes the GTS descriptor from the beacon once the associated node acknowledges its reception. Indeed, nodes that do not support DEEP can still follow the 802.15.4 standard in keeping the unacknowledged GTS descriptor in all beacons. This makes DEEP backward compatible. DEEP can effectively reduce the size of the communicated beacons and thus the power consumed to process them at the receiving nodes is significantly reduced. In [4], the authors tackle the problem of packet collisions encountered with beacon-enabled 802.15.4. Two novel hashing channel selection mechanisms are introduced to reduce the amount of collisions during the slotted CSMA-CA period. These mechanisms are random and enhanced linear probing and their details can be found in [4]. The authors demonstrate through extensive simulations that their proposed mechanisms outperform the original 802.15.4 standard in terms of throughput, utilization, average delay, and packet drop-off. This study lacks any examination of the power requirements of the proposed hashing schemes. The latter is crucial in WSNs. In [5], the authors propose modifications to beacon-enabled 802.15.4 to improve its performance in terms of latency and reliability, and therefore, make it conform to the requirements of industrial applications. Towards that, the authors propose three schemes 1) swap the positions of CAP and CFP in the superframe and allow failed GTS frames to be retransmitted in during the CAP, 2) in the original 802.15.4, allow the retransmission of GTS frames in the CAP of the following superframe, and 3) in the original 802.15.4, allow the retransmission of GTS frames in the CAP of the current frame. Through a simulation study, the authors point out the gains they can achieve with their modifications over the original 802.15.4. The main drawback of the schemes in [5] is that the modifications proposed require significant changes in 802.15.4 and changes its nature. As a result, a thorough study will be needed to examine if the performance of the different aspects of the original standard is not negatively affected. Compared to the contributions above, our proposed extension requires minor modification to the behavior of the nodes in an 802.15.4-based WSN. In particular, we restrict
the frequency at which nodes can access the communication channel. This is done while preserving the basic operation of the original standard without introducing any new control messages or changing the superframe format. Furthermore, we do not introduce any new computational burden upon sensor nodes.
IV.
SIMULATIONS
To study the performance of the network with the introduced SB state, we conduct simulations to study the performance in terms of fairness, channel utilization, reliability, and most importantly power consumption. We wrote a simulator using C language. The network under simulation is of a peer-to-peer topology. The simulation parameters are stated in Table I (we use some of the parameters used in [8]). We assume no CFP or inactive periods. That is, nodes always follow the standard slotted CSMA-CA approach, with the introduced SB state, to access the communication medium. The data traffic is assumed to be saturated. That is, each node always has data to be sent out. Also, the traffic is assumed to be acknowledged. In Fig. 2, we show a flow chart for the slotted CSMA-CA algorithm. The flow chart is a modified version of the chart illustrated in [1], in which the un-slotted CSMA-CA is also included. In the same figure, we highlight in red the newly added SB state. The flow chart clearly reflects our extension in which a node is forced to standby after a successful transmission and before commencing its trials to access the medium again. We now show and discuss the data collected to study each of the performance parameters mentioned earlier. A. Fairness It is essential to preserve fairness among nodes. That is, each node should have an equal probability of accessing the
communication medium. We compute the fairness index using Jain’s formula [6]:
∑ ∑ where, N is the number of contending nodes, and xi is the medium share of the ith node. A fairness index of 1 implies that the protocol used is allowing each node in the network an equal share of medium access. However, as the fairness index decreases towards zero, it implies that only a portion of the nodes is getting more opportunities to access the medium than other nodes. In Fig. 3, for network sizes ranging from 10 to 50 nodes, we show the performance in terms of fairness under different values of LSB. The figure clearly shows that we are almost always at a fairness index of 1, indicating that our extension to the 802.15.4 standard is achieving a fair treatment of the nodes in the network. B. Channel Utilization It is essential to examine the behavior of the extended 802.15.4 in terms of wireless channel usage. We need to make sure that the proposed extension can achieve at least the same level of channel utilization (U) as the original 802.15.4 standard. We define U to be , that is, the ratio of the packet length (L) to the total period of time (D) spent starting from sensing the channel till the ACK packet is received back. It is essential to note that D takes into account the transmission time (Ls), the time wasted due to j collisions (jLc), and the time spent during the backoff stages (DBO). As we introduce the SB state, we need to account, in our computation of U, for this extra time of inactivity. In other words, once all nodes finish a successful transmission of a packet, and assuming that LSB is set to a long duration that exceeds Ls+jLc+DBO, we will eventually reach a stage where all nodes are in the SB state. At that stage, the communication medium will be idle, and therefore, the SB will be contributing to the underutilization of the medium. In Fig. 4, for network sizes ranging from 10 to 50 nodes, we show the performance in terms of U under different values of LSB. Interestingly enough, we observe that U shows an increasing trend till it reaches a maximum value (for example, for a network of 10 nodes, U reaches a value of 57.69% at an LSB of 150 timeslots, before declining as we increase the value of LSB. The reason behind the increasing trend is that forcing nodes to go to the SB state is playing a significant role in reducing the number of contending nodes and thus reducing the probability of packet collisions.
Therefore, while LSB is small, nodes are getting better opportunities of using the medium in order to successfully transmit their data. However, as we keep increasing LSB, as we mentioned earlier, we are actually forcing nodes to sleep for extended periods of time after each successful transmission. Once all nodes are in the SB state, the medium may remain idle for a long duration before one node starts the channel sensing again. The most important conslusion we gain from Fig. 4 is that different sizes of WSNs are reaching a maximum (that is, optimum) value of U under different values of LSB. Therefore, we can tune the LSB’s value according to the newtork size such that an optimum U is achieved. Based on the latter knowledge, we can draw the graph in Fig. 5. This figure determines, for a given network size, the optimal value of LSB that can guarantee the maximum value of U. C. Reliability As in [7], we define reliability (R) as the probability of receiving a packet successfully. In Fig. 6 we show the behavior of R under different LSB values and for different network sizes. Apparently, as LSB increases, R tends to reach, and remain at, the value of 1. This is due to that as node standby for longer periods, they actually contribute to excessive reduction in the probability of collision and thus more packets are reaching their destinations successfully. Fig. 6 also shows that as the size of the network gets bigger, R keeps at a lower value before it starts its rise towards the 1 value. This is due to fact that as the number of nodes increases in the network, packet collisions increase and therefore we will need to force nodes to standby for longer periods of time, giving better chances of successful transmission for the awaken nodes. In Fig. 7 we draw the values of R achieved at the optimum LSB values found in subsection B above. In this figure, we compare these values of R with the reliability achieved under the original 802.15.4 standard. Both curves are shown at different network sizes. We can clearly see that with the SB state we are outperforming the original standard and a significant improvement in R is achieved. D. Power Consumption In Fig.8 we illustrate the effect of introducing the SB state on the average power consumption (P) (which is the sum of the power consumed during transmission, reception, sleep, CCA, and collisions). For different network sizes, we show P as the value of LSB increases. This figure clearly shows the expected behavior of achieving more power savings as we force nodes to standby for longer periods. Also apparent in
the same figure that networks constituted by 10 nodes suffer from relatively high P at lower values of LSB. This can be understood if we notice that as we decrease N, nodes get better chances of successfully transmitting their packets (i.e., packet collisions are low). Therefore, since LSB is low, nodes will be actively involved in accessing the communication channel to send packets and this contributes to higher P. In Fig. 9, we use the optimum values of LSB (from sub-section B) to find the power consumption experienced with them. We compare the latter power consumption with that experienced in under the original 802.15.4 standard. The graph clearly shows the savings we can achieve by introducing the SB state in the 802.15.4 standard. V.
CONCLUSION
In this paper we propose an extension to the IEEE 802.15.4 standard that is ought to improve its performance in terms of power conservation, reliability, and channel utilization while preserving fairness among nodes in the WSN. The extension introduces a new standby state that each node should enter upon completing a successful packet transmission. We anticipated that the new state not only allows nodes to save more power, but also reduces the contention among nodes, which reflects in improving the performance parameters mentioned above. We built a simulator based on C language and conducted extensive simulations on a peer-to-peer WSN to study the performance of our extended 802.15.4. The simulations confirmed our expectations and revealed that, given network sizes, nodes can be tuned at optimal standby durations (LSB) such that maximum channel utilization can be achieved. At these LSB values, we can outperform the original 802.15.4 standard in terms of channel utilization, reliability and power consumption while preserving fairness among nodes. VI.
REFERENCES
[1] IEEE Std 802.15.4-2006, September, Part 15.4: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (WPANs). Available online at: http://standards.ieee.org/getieee802/download/802.15.42006.pdf (Last accessed in Oct. 2010). [2] Y. Gadallah and M. Jaafari, “A Reliable Energy-Efficient 802.15.4-Based MAC Protocol for Wireless Sensor Networks,” Proceedings of IEEE Wireless Communications and Networking Conference (WCNC’10), Sydney, Australia, Apr. 2010. [3] M. Valero, A. Bourgeois, and R. Beyah, “DEEP: A Deployable Energy Efficient 802.15.4 MAC Protocol for Sensor Networks,” Proceedings of IEEE International
Conference on Communications (ICC’10), Cape Town, South Africa, May 2010. [4] B.-Y. Shih, C.-Y. Chen, C.-H. Shih, and J.-Y. Tseng, “The Development of Enhancing Mechanisms for Improving the Performance of IEEE 802.15.4,” International Journal of the Physical Sciences, vol. 5 (6), pp. 884-897, Jun. 2010. [5] G. Bhatti, A. Mehta, Z. Sahinoglu, J. Zhang, and R. Viswanathan, “Modified Beacon-Enabled IEEE 802.15.4 MAC for Lower Latency,” Proceedings of IEEE Global Telecommunications Conference (GLOBECOM’08), pp. 876-880, Nov./Dec. 2008. [6] R. Jain, D. Chiu, and W. Hawe, “A Quantitative Measure of Fairness and Discrimination for Resource Allocation in Shared Computer Systems”, DEC-TR-301, Sept. 26th, 1984. [7] P. Park, P. D. Marco, C. Fischione, and K. H. Johansson, "Adaptive IEEE 802.15.4 protocol for reliable and timely communications," KTH, Tech. Rep. TRITA-EE 2009:054, 2009. [8] S. Pollin, M. Ergen, S. C. Ergen, B. Bougard, L. Van der Perre, I. Moerman, A. Bahai, P. Varaiya, and F. Catthoor, “Performance Analysis of Slotted Carrier Sense IEEE 802.15.4 Medium Access Layer,” IEEE Transactions on Wireless Communications, vol. 7, no. 9, pp. 3359-3371, Sept. 2008.
Fig. 3. Fairness under different values of LSB for different network sizes.
Fig. 4. Channel Utilization under different values of LSB for different network sizes.
Fig. 8. Average Power Consumption under different values of LSB for different network sizes.
Fig. 5. Optimal values of LSB to achieve maximum U, given the network’s sizes.
Fig. 9. Average power consumption experienced at the optimal LSB values for different network sizes.
Fig. 6. Reliability under different values of LSB for different network sizes.
Fig. 7. Reliability achieved at the optimal LSB values, for different network sizes.
Table I. Simulation Parameters Rx 40 Power Tx 30 Consumed CCA 40 (mW) Sleep 0.8 1 timeslot 0.32 ms Packet Length (L) 14 timeslots Durations 2 timeslots ACK Packet Length (LACK) Simulation Time 320 s 802.15.4 macMaxCSMABackoffs 5 Parameter aMinBE 3 Settings aMaxBE 5
