From versatility to auto-adaptation of the medium access control in wireless sensor networks

https://doi.org/10.1016/j.jpdc.2010.10.001Get rights and content

Abstract

Recent deployments of wireless sensor networks have targeted challenging monitoring and surveillance applications. The medium access control being the main source of energy wastage, energy-efficiency has always been kept in mind while designing the communication stack embedded in spread sensors. Especially, versatile protocols have emerged to offer a suitable solution over multiple deployment characteristics. In this study, we observe to what extent versatility applies to dynamic scenarios in which communications do not respect specific communication paradigms. We first provide a performance evaluation of two well-reputed versatile protocols (B-MAC (Polastre et al., 2004 [17]) and X-MAC (Buettner et al., 2006 [4])) under the conditions of such a scenario. The obtained results convinced us to propose more than versatility and pre-configured solutions, that is auto-adaptation. We then introduce the main contribution of this paper, an auto-adaptive algorithm that allows one to adjust the previously mentioned protocols while the network is operating. We analyze to what extent it outperforms the previously obtained results.

Introduction

After near a decade of theoretical research contributions, Wireless Sensor Networks (WSN) have now become reality. More and more people have been deploying sensors for various reasons, either for the sake of concrete applications or merely for deploying and experimenting protocols under real conditions (radio environment, mobility of nodes, desynchronized clocks, etc.) [16], [22], [12], [3], [6]. Sensing the environment (temperature, humidity, light intensity, noise, acceleration) in the least intrusive fashion is indeed the main advantage brought by WSN as data are silently collected and reported to the sink station. This reporting can be achieved either during regular periods or in real-time (upon an event for instance). We will detail in Section 2.1 the multiple data collection schemes that exist, and explain the limitations of their associated deployments. This will lead us to define a likely scenario that could prefigure the future of the WSN deployments.

In any case and to the best of our knowledge, no deployment has ever taken place without considering the need for energy efficiency. With wireless communications being the first source of energy depletion, efforts have focused on the optimization of the communication stack, especially at the routing and medium access control (MAC) layers. The most commonly adopted routing policy seems to be a tree whose branches are followed by data to reach the sink [22], [3], [6]. This sink-rooted tree is the direct consequence of the N-to-1 communication interactions in WSN. Regarding the MAC, choices are not that clear. In Section 2.2, we will explain why current MAC protocols may not be suitable when confronted to the scenario we defined.

We decided to focus our attention on sampling MAC protocols rather than on time-slotted MAC solutions in which nodes are organized around a common schedule (S-MAC [24] or IEEE 802.15.4 [11]). In this latter set of solutions, time is divided into slots distributed among the nodes, which agree with one another to use such slots to send or receive data, or to power off the radio. To our minds, this kind of scheme brings up too many assumptions and limitations to be considered once addressing a set of applications as large as possible. The imposed time synchronization, the bounded number of time-slots in the communication window or the need for a coordinator make them unsuitable and not scalable to dynamic scenarios with a large number of nodes. For instance, slotted protocols can hardly integrate mobile sensors in their communication scheduling algorithms. Indeed, they have to compute and distribute again the time-slots every time a new sensor wishes to participate in the communication.

The design of sampling protocols confers to them two key advantages. First, no strict time synchronization is required among the nodes that use preamble sampling techniques. This removes the overhead of a synchronization protocol. Sampling protocols do not suffer from a bounded number of time-slots either. Theoretically, every node is able to use the medium. For all these reasons, we will concentrate on sampling protocols in the remainder of this paper. As we will detail in Section 2.3, versatility has recently been brought in the front scene with protocols such as B-MAC [17], thus allowing the pre-configuration of the same MAC layer for various kinds of applications and communication schemes. Yet, auto-adaption once nodes have been deployed has not been much discussed so far.

As we investigated the performances of two well-reputed MAC protocols [17], [4] in Section 3, we felt the need to propose more than versatility and pre-configured solutions, that is auto-adaptation. Our solution detailed in Section 4 uses the versatile features of the protocols to make them truly auto-adaptive. It was kept as simple as possible in order not to participate in increasing the complexity emerging from an already vast set of proposed MAC contributions. This first step toward auto-adaption of energy-efficient MAC solutions is promising as the obtained results show that X-MAC [4] provides better performances when coupled with our algorithm. We end this paper by sharing in Section 5 our envisioned improvements and steps that could be made in order to move from versatility to auto-adaptive medium access control solutions.

Section snippets

New challenges in WSN deployments

During the last decade, numbers of WSN deployments have been reported in the literature [18]. The large majority can usually be classified into two distinct categories of applications. The first one serves the purpose of continuous observations such as glacier [3] or habitat [16] monitoring. The sensor nodes report their measured values toward a sink in a time-driven fashion. The second category operates with a transient report model, typically when a specific event is observed in the network.

Operating a pre-configured MAC on a dynamic scenario

We have implemented the dynamic scenario presented in Section 2.1 and evaluated the performances of B-MAC and X-MAC when different preamble lengths are used with this very scenario.

Toward an auto-adaptive MAC layer

Regarding all previously stated drawbacks of a pre-configured MAC design, we now propose a solution to dynamically adapt the LPL mechanism. Once again, we insist on how different versatility and auto-adaptation are. Here, interfaces available in a MAC layer such as B-MAC enable versatility. We aim at introducing auto-adaptation, that is using this versatility feature to automatically tune the LPL mechanism and optimize energy savings. For the sake of clarity, the preamble length is later

Envisioned optimizations

We have briefly introduced in the previous section several possible optimizations for our solution. We now detail three possible enhancements that could further improve our algorithm by increasing the proportion of small preambles sent in the network. Along with these proposals, we also expose preliminary results that will help to understand the possible benefits.

Conclusion

A survey of WSN deployments [13] has revealed that so far, most of them were operated at small-scale, for a short period of time and with a knowledge a priori of the data collection scheme. They usually have been managed at the MAC layer by simple protocols that were however hardly usable in different scenarios. In parallel, the research community has released a considerable number of protocols destined to WSN. Especially, the B-MAC and X-MAC sampling protocols propose a set of interfaces to

Romain Kuntz received Ph.D. in computer science in 2010 from Strasbourg University, France. He is a member of the Network Research team at the LSIIT Laboratory (UMR CNRS 7005). After receiving his M.Sc. in computer science in 2004, he has worked for 3 years as a research engineer in Keio University and the University of Tokyo in Japan. His research interests include multihoming and mobility management over IPv6 networks, and MAC protocols for wireless sensor networks.

References (25)

  • M. Ali et al.

    Medium access control issues in sensor networks

    ACM SIGCOMM Computer Communication Review

    (2006)
  • J. Allred et al.

    Sensorflock: an airborne wireless sensor network of micro-air vehicles

  • G. Barrenetxea et al.

    Sensorscope: out-of-the-box environmental monitoring

  • M. Buettner et al.

    X-mac: a short preamble mac protocol for duty-cycled wireless sensor networks

  • R. Cardell-Oliver et al.

    Field testing a wireless sensor network for reactive environmental monitoring

  • M. Ceriotti et al.

    Monitoring heritage buildings with wireless sensor networks: the torre aquila deployment

  • Chipcon cc1100 chipset....
  • Crossbow technology....
  • K. Dantu, M. Rahimi, H. Shah, S. Babel, A. Dhariwal, G.S. Sukhatme, Robomote: enabling mobility in sensor networks in:...
  • I. Demirkol et al.

    Mac protocols for wireless sensor networks: a survey

    IEEE Communications Magazine

    (2006)
  • IEEE, Part 15.4: wireless medium access control (mac) and physical layer (phy) specifications for low-rate wireless...
  • S. Kim et al.

    Health monitoring of civil infrastructures using wireless sensor networks

  • Cited by (0)

    Romain Kuntz received Ph.D. in computer science in 2010 from Strasbourg University, France. He is a member of the Network Research team at the LSIIT Laboratory (UMR CNRS 7005). After receiving his M.Sc. in computer science in 2004, he has worked for 3 years as a research engineer in Keio University and the University of Tokyo in Japan. His research interests include multihoming and mobility management over IPv6 networks, and MAC protocols for wireless sensor networks.

    Antoine Gallais received M.Sc. (2004) and Ph.D. (2007) Degrees in computer science from the University of Lille, France. In 2008, he joined the Image Sciences, Computer Sciences and Remote Sensing Lab (LSIIT), located in the University of Strasbourg, France, where he is currently an associate professor. His main research interests lie in the areas of wireless sensor and mobile ad hoc networking, mobility management and network security. He is especially interested in localized and distributed solutions for sensing coverage by connected sets, energy-efficient medium access control and routing in wireless sensor networks. He is involved in the french national project Senslab, whose goal is to open a large-scale wireless sensor network testbed, for the needs of research and teaching communities. He is also investigating new sensor communication solutions for home automation and tele-surveillance for remote medical care purposes. Up to date information can be later found at http://clarinet.u-strasbg.fr/~gallais.

    Thomas Noël is a professor at Strasbourg University, France. He is in charge of all wireless IP-mobility activities in the Network Research team at the LSIIT laboratory (UMR CNRS 7005). His research interests include multicast routing for mobile devices, multihomed devices, optimization of handovers, energy consumption policies on mobile handheld devices, and sensor networks.

    View full text