Table of Contents Author Guidelines Submit a Manuscript
Journal of Applied Mathematics
Volume 2014 (2014), Article ID 138571, 12 pages
http://dx.doi.org/10.1155/2014/138571
Research Article

Taxonomy and Evaluations of Low-Power Listening Protocols for Machine-to-Machine Networks

Department of Embedded Systems Engineering, Incheon National University, Incheon 402-772, Republic of Korea

Received 2 April 2014; Accepted 4 June 2014; Published 8 July 2014

Academic Editor: Young-Sik Jeong

Copyright © 2014 Kwang-il Hwang and Sung-Hyun Yoon. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Even though a lot of research has made significant contributions to advances in sensor networks, sensor network protocols, which have different characteristics according to the target application, might confuse machine-to-machine (M2M) network designers when they choose the protocol most suitable for their specific applications. Therefore, this paper provides a well-defined taxonomy of low-power listening protocols by examining in detail the existing low-power sensor network protocols and evaluation results. It will also be very useful for helping M2M designers understand specific features of low-power media access control protocols as they design new M2M networks.

1. Introduction

Machine-to-machine (M2M) networks enable creation of the Internet of Things, which interconnects via the Internet physical things equipped with various sensors and actuators. Mitsui et al. [1] presented various M2M applications based on sensor network technologies. A typical M2M architecture is basically composed of three domains: the server, the Internet, and the sensors. In particular, the sensor domain is the most important, aggregating data from physical sensors and accessing the Internet via 3G or 4G M2M gateways. Like a sensor network, an M2M sensor domain requires well-structured and energy-efficient network protocols among distributed sensors using short range communications. Much research has already been conducted on sensor network protocols [2], making significant contributions towards advances in automated sensor networks [35]. However, having too many sensor network protocols causes confusion for M2M designers as they choose the protocol most suitable for their specific applications. Furthermore, most of the literature on sensor network protocols is too theoretical, requires a lot of specific assumptions, and is not easy to apply to practical M2M sensor domains.

Sensor media access control (MAC) protocols can be categorized into random-based, slot (schedule-) based, time division multiple access- (TDMA-) based, random/TDMA hybrids, and low-power listening (LPL) methods. In particular, LPL-based MAC protocols can be considered the most suitable type for M2M sensor domains, because they provide a low duty cycle and low implementation complexity. Therefore, there has been substantial research on LPL protocols. Each one shows different characteristics and operations, as described in Table 1. Therefore, this paper aims to provide a well-defined taxonomy of low-power listening protocols by examining in detail the existing low-power sensor network protocols, introducing an M2M communication model and then evaluating performance with respect to data aggregation time and energy consumption in terms of an M2M communication model.

tab1
Table 1: LPL MAC protocols.

The remainder of this paper is organized as follows. A taxonomy of LPL protocols is presented in Section 2. Section 3 analyzes each LPL protocol in terms of an M2M communications model. Section 4 summarizes numerical results and Section 5 provides concluding remarks.

2. A Taxonomy of Low-Power Listening Protocols

2.1. Trigger Source (Preamble versus Packet)

The main idea of LPL is to asynchronously trigger a receiver that is alternating between wake-up and sleep states to detect a wake-up signal from a sender. Therefore, receivers can save much more energy by removing idle listening periods. Some protocols, such as B-MAC, WISE-MAC, and X-MAC, use a preamble as a trigger source. On the other hand, SpeckMAC, RI-MAC, BoX-MAC, MX-MAC, and A-MAC trigger receivers by transmitting a consecutive packet. More specifically, RI-MAC, MX-MAC, A-MAC, and SPEC-MAC-D utilize a data packet for the trigger, and SpeckMAC-B, BoX-MAC-1, and BoX-MAC-2 utilize short wake-up packets before data transmission.

2.2. Initiation Method (Receiver-Initiated versus Source-Initiated)

LPL protocols can also be categorized into source-initiated and receiver-initiated methods, according to which one begins the transmission request. RI-MAC and A-MAC are receiver-initiated protocols but the rest of the protocols are source-initiated protocols.

2.3. Adaptivity (Adaptive versus Deterministic)

B-MAC, SpeckMAC, RI-MAC, A-MAC, and BoX-MAC-1 always transmit triggering signals for predetermined fixed duration, but some protocols, such as WISE-MAC, X-MAC, MX-MAC, and BoX-MAC-2, transmit variable triggering signals depending on when a receiver is triggered.

2.4. Schedule (Schedule versus Nonschedule)

To reduce data pending time more, some protocols, such as WISE-MAC and MX-MAC, use schedule-based triggering by exchanging wake-up time information among neighbors.

3. M2M Communication Model

In this section, an M2M communication model is presented, and then each LPL protocol is analyzed in terms of the M2M model.

3.1. System Model

Generally, M2M is composed of a concentrator, which is a centralized device to connect the M2M sensor domain to the Internet, and M2M devices, which are equipped with various sensors or actuators. In an M2M sensor domain, devices form either a star or a peer-to-peer topology for multihop communications. Data from each device are aggregated in the concentrator and transmitted to a corresponding server via the Internet. To consider a practical M2M system, each protocol and algorithm should be able to execute their tasks with off-the-shelf radio frequency (RF) modems (TI CC430, CC2420, RadioPulse MG2400, etc.) and MCUs.

3.2. Data Model

The most popular data models for M2M are the periodic report model and the request-oriented model. In the periodic report model, each device transmits data to a concentrator periodically, and the model is generally used for unidirectional data aggregation. By contrast, the request-oriented model allows bidirectional communication between the concentrator and devices. In the data model, a server (user) can request a concentrator to aggregate real-time sensor data in the sensor domain. The concentrator also triggers and transmits server requests to the devices. Each device replies to the concentrator, and the responses from devices are aggregated in the concentrator and transmitted to the server.

3.3. Energy Model

For M2 M networks, energy conservation is one of the most critical challenges, as it is in sensor networks. It is important to note that in order to save energy, each device should remain active only for required duration, and the rest of the time should go to sleep. Therefore, when calculating the energy consumption of each device, we need to know the total active duration, (i) and the total sleep duration, (ii) in a request interval, . By using (i) and (ii), the energy consumption for each device can be expressed as follows:

4. Numerical Analysis

Now, we numerically analyze each LPL protocol in terms of M2M communication models. In particular, we focus on data aggregation time, which is the total time required to aggregate data from all devices with respect to a request. Table 2 presents notations used for our numerical analysis.

tab2
Table 2: Notations.
4.1. B-MAC

B-MAC is a representative LPL protocol utilizing a preamble for the receiver trigger. As shown in Figure 1, each device repeats a short time wake-up for to detect the preamble transmission and then sleeps for , per . A sender that wants to send data first transmits a long preamble for to trigger the receiver that is performing periodic preamble sensing (PPS) before data transmission. Each preamble transmission can be detected by all devices within communication range of a sender, and all nodes that detect the preamble transmission, as well as the intended receiver, have to remain active for , until the preamble transmission ends.

138571.fig.001
Figure 1: B-MAC.
4.1.1. Periodic Report

Since each device should send its data to the concentrator on the predetermined schedule, the report time of each device is as follows: Therefore, the total report time of nodes is

4.1.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is In particular, since B-MAC is capable of triggering all nodes with a single preamble transmission, all the devices can listen to the request message following the preamble. Therefore, the total aggregation time of nodes per request is

4.2. WISE-MAC

Unlike B-MAC, which transmits a long preamble, WISE-MAC aims to save more energy by transmitting a shorter preamble. To achieve this, each sender manages a schedule table in which all its neighbors’ PPS schedules are stored. Therefore, if a sender does not know the PPS schedule of a receiver, the sender must transmit a long preamble, as in B-MAC, but otherwise, the sender can transmit a minimum preamble for to trigger the receiver, as shown in Figure 2.

138571.fig.002
Figure 2: WISE-MAC.
4.2.1. Periodic Report

The report time of a WISE-MAC device is calculated as follows: Therefore, the total report time of nodes is

4.2.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is as follows: And, like B-MAC, the WISE-MAC sender can trigger all devices with a single long preamble, so the total aggregation time of nodes per request is

4.3. X-MAC

As shown in Figure 3, X-MAC utilizes an early acknowledgment (ACK) and short preamble to reduce energy waste from transmitting a long preamble, as in B-MAC and WISE-MAC in the worst case scenario. In addition, the short preamble represents a destination ID, and a gap, , between short preambles is used to receive the ACK of the receiver.

138571.fig.003
Figure 3: X-MAC.
4.3.1. Periodic Report

The report time of an X-MAC device is as follows: where is a maximum time to trigger the receiver and .

Therefore, the total report time of nodes is

4.3.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is And, unlike B-MAC and WISE-MAC, X-MAC cannot trigger all devices with a single preamble transmission. Therefore, the request of the concentrator must be transmitted as many times as the number of devices. So the total aggregation time of nodes per request is

4.4. SpeckMAC-B

Instead of a preamble transmission, SpeckMAC-B transmits consecutive wake-up packets to trigger devices performing periodic wake-up-signal sensing (PWS), as shown in Figure 4. A wake-up packet contains a destination ID and a time stamp, which represents data packet transmission time information. Therefore, a device that listens to a wake-up packet during PWS goes to sleep until the beginning of data transmission wakes up and then receives data from the sender. Devices that listen to a wake-up packet but that are not the intended receiver go to sleep and continue to perform PWS.

138571.fig.004
Figure 4: SpeckMAC-B.
4.4.1. Periodic Report

The report time of a SpeckMAC-B device is as follows: where is a maximum time to trigger the receiver and .

Therefore, the total report time of nodes is

4.4.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is And, like B-MAC or WISE-MAC, a single request packet can trigger all devices, so the total aggregation time of nodes per request is

4.5. SpeckMAC-D

Instead of the wake-up packet transmission used in SpeckMAC-B, SpeckMAC-D enables fast data reception by utilizing consecutive data packets. Each SpeckMAC-D device performs periodic frame sensing (PFS) for to receive a data frame, as shown in Figure 5.

138571.fig.005
Figure 5: SpeckMAC-D.
4.5.1. Periodic Report

The report time of a SpeckMAC-D device is as follows: where is a maximum time to trigger the receiver and .

Therefore, the total report time of nodes is

4.5.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is And is equal to . So the total aggregation time of nodes per request is

4.6. RI-MAC

RI-MAC is a representative receiver-initiated LPL protocol. Each RI-MAC device basically performs periodic beacon sending (PBS). A sender first switches to reception (RX) mode and waits to receive the beacon of a corresponding receiver. As soon as a corresponding beacon is received, the sender transmits data and then goes back to PBS. The receiver of the data acknowledges a beacon, as shown in Figure 6.

138571.fig.006
Figure 6: RI-MAC.
4.6.1. Periodic Report

The report time of an RI-MAC device is as follows: Therefore, the total report time of nodes is

4.6.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is And is equal to . So the total aggregation time of nodes per request is

4.7. BoX-MAC-1

As shown in Figure 7, BoX-MAC-1 is one of the packet-based LPL protocols, and (to wait for ACK of the receiver) is followed by consecutive wake-up packets. Then, on reception of the ACK, the data are transmitted.

138571.fig.007
Figure 7: BoX-MAC-1.
4.7.1. Periodic Report

The report time of a BoX-MAC-1 device is as follows: where is a maximum time to trigger the receiver and .

Therefore, the total report time of nodes is

4.7.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is And since data transmission is started only when an ACK is received, the request of the concentrator must be transmitted as many times as the number of devices. So the total aggregation time of nodes per request is as follows:

4.8. BoX-MAC-2

BoX-MAC-2 is one of the wake-up, packet-based LPL protocols. However, unlike BoX-MAC-1 or SpeckMAC-B utilizing consecutive wake-packet transmissions, a sender waits for ACK from the receiver for , per wake-up transmission, as shown in Figure 8. Therefore, a sender repeats wake-up packet transmission and RX for ACK until receiving the ACK.

138571.fig.008
Figure 8: BOX-MAC-2.
4.8.1. Periodic Report

The report time of a BoX-MAC-2 device is as follows: where is a maximum time to trigger the receiver and .

Therefore, the total report time of nodes is

4.8.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is And as in BoX-MAC-1, since data transmission is started only when ACK is received, the request of the concentrator must be transmitted as many times as the number of devices. So the total aggregation time of nodes per request is

4.9. MX-MAC

MX-MAC is one of the packet-based LPL protocols and aims to reduce additional handshakes to trigger the receiver. Each MX-MAC device performs PFS, and a sender transmits data after waiting for and then switches to RX to listen for ACK for , as shown in Figure 9.

138571.fig.009
Figure 9: MX-MAC.
4.9.1. Periodic Report

The report time of an MX-MAC device is as follows: where is a maximum time to trigger the receiver and .

Therefore, the total report time of nodes is

4.9.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is is equal to , and each request is paired with each response. So the total aggregation time of nodes per request is

4.10. A-MAC

A-MAC is one of the receiver-initiated LPL protocols, like RI-MAC. However, it enhances data reliability by adding early ACK transmission before data transmission. A sender goes into RX to wait and listen for a beacon from the receiver. Upon reception of the corresponding beacon, the sender transmits an early ACK to notify the receiver that data transmission follows, as shown in Figure 10.

138571.fig.0010
Figure 10: A-MAC.
4.10.1. Periodic Report

The report time of an MX-MAC device is as follows: Therefore, the total report time of nodes is

4.10.2. Request-Oriented

The required time for a concentrator to transmit its request to devices is is equal to , and each request is paired with each response. So the total aggregation time of nodes per request is

5. Summary and Concluding Remarks

Based on the taxonomy presented in Section 2 and the numerical analysis presented in Section 4, Table 3 presents a summary of the taxonomy and evaluation results regarding data aggregation time and energy consumption in terms of M2M communication models. In addition, protocol complexity is evaluated in terms of time synchronization requirements, memory usage, and ability to be implemented with an off-the-shelf RF modem and MCU.

tab3
Table 3: Summary of taxonomy and evaluations.

First, in terms of data aggregation time, without regarding wake-up source type, while adaptive and schedule-based protocols such as WISE-MAC and MX-MAC show fast data aggregation time, preamble-based or receiver-initiated protocols like B-MAC, RI-MAC, and A-MAC present long aggregation times. This is because adaptive LPL protocols are capable of coping with a receiver’s reaction through feedback during wake-up duration, compared with deterministic protocols utilizing fixed-size wake-up duration without regard to the receiver’s reaction.

In terms of energy efficiency, while preamble-based protocols, such as B-MAC, WISE-MAC, and X-MAC present superior energy efficiency, data packet-based LPL protocols like BoX-MAC and MX-MAC present high energy consumption. Since preamble detection duration is considerably shorter than the data reception duration, the preamble-based protocols can operate with a very short duty cycle.

In terms of protocol complexity, deterministic or receiver-initiated protocols have relatively low complexity, whereas adaptive and schedule-based protocols, such as WISE-MAC, SpeckMAC-B, and MX-MAC, have high complexity because they require tight time synchronization and management for neighbors’ PPS times. In addition, X-MAC (which transmits a short preamble in which ID information is contained) is not possible to implement with an off-the-shelf RF modem.

Lastly, we expect the summarized taxonomy will provide a useful guideline for understanding the specific features of LPL protocols and for designing a new M2M network.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

Acknowledgments

This work was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) and funded by the Ministry of Education, Science and Technology (2012R1A1A2041271).

References

  1. H. Mitsui, H. Kambe, and H. Koizumi, “Student experiments for learning basic M2M technologies by implementing sensor network systems,” in Proceedings of the 9th International Conference on Information Technology Based Higher Education and Training (ITHET '10), pp. 268–275, Cappadocia, Turkey, April-May 2010. View at Publisher · View at Google Scholar · View at Scopus
  2. P. Huang, L. Xiao, S. Soltani, M. W. Mutka, and N. Xi, “The evolution of MAC protocols in wireless sensor networks: a survey,” IEEE Communications Surveys & Tutorials, vol. 15, no. 1, pp. 101–120, 2013. View at Publisher · View at Google Scholar · View at Scopus
  3. J. Liu and S. H. Chung, “An efficient load balancing scheme for multi-gateways in wireless mesh networks,” Journal of Information Processing Systems, vol. 9, no. 3, pp. 365–378, 2013. View at Google Scholar
  4. A. Sinha and D. K. Lobiyal, “Performance evaluation of data aggregation for cluster-based wireless sensor network,” Human-Centric Computing and Information Sciences, vol. 3, no. 13, pp. 1–17, 2013. View at Publisher · View at Google Scholar
  5. M. Yoon, Y. K. Kim, and J. W. Chang, “An energy-efficient routing protocol using message success rate in wireless sensor networks,” Journal of Convergence, vol. 4, no. 1, 2013. View at Google Scholar
  6. J. Polastre, J. Hill, and D. Culler, “Versatile low power media access for wireless sensor networks,” in Proceedings of the 2nd International Conference on Embedded Networked Sensor Systems (SenSys '04), pp. 95–107, New York, NY, USA, November 2004. View at Publisher · View at Google Scholar · View at Scopus
  7. A. El-Hoiydi and J. Decotignie, “WiseMAC: an ultra low power MAC protocol for the downlink of infrastructure wireless sensor networks,” in Proceedings of the 9th International Symposium on Computers and Communications (ISCC '04), pp. 244–251, June-July 2004. View at Publisher · View at Google Scholar · View at Scopus
  8. M. Buettner, G. V. Yee, E. Anderson, and R. Han, “X-MAC: a short preamble MAC protocol for duty-cycled wireless sensor networks,” in Proceedings of the 4th International Conference on Embedded Networked Sensor Systems (SenSys '06), pp. 307–320, New York, NY, USA, November 2006. View at Publisher · View at Google Scholar · View at Scopus
  9. K.-J. Wong and D. Arvind, “SpeckMAC: low-power decentralised MAC protocols for low data rate transmissions in specknets,” in Proceedings of the 2nd International Workshop on Multi-hop Ad Hoc Networks: From Theory to Reality (REALMAN '06), pp. 71–78, New York, NY, USA, May 2006. View at Publisher · View at Google Scholar · View at Scopus
  10. Y. Sun, O. Gurewitz, and D. B. Johnson, “RI-MAC: a receiver-initiated asynchronous duty cycle MAC protocol for dynamic traffic loads in wireless sensor networks,” in Proceedings of the 6th ACM Conference on Embedded Networked Sensor Systems (SenSys '08), pp. 1–14, New York, NY, USA, November 2008. View at Publisher · View at Google Scholar · View at Scopus
  11. D. Moss and P. Levis, “BoX-MAC: exploiting physical and link layer boundaries in low-power networking,” Tech. Rep. SING-08-00, 2008. View at Google Scholar
  12. C. J. Merlin and W. B. Heinzelman, “Schedule adaptation of low-power-listening protocols for wireless sensor networks,” IEEE Transactions on Mobile Computing, vol. 9, no. 5, pp. 672–685, 2010. View at Publisher · View at Google Scholar · View at Scopus
  13. P. Dutta, S. Dawson-Haggerty, Y. Chen, C. M. Liang, and A. Terzis, “Design and evaluation of a versatile and efficient receiver-initiated link layer for low-power wireless,” ACM Transactions on Sensor Networks, vol. 8, no. 3, 2012. View at Google Scholar