Sensor Media Access Control
Sensor Media Access Control [ ˈsɛnsə ˈmiːdiə ˈæksɛs kənˈtɹəʊl ] ( S-MAC , " Sensor Media Access Control ") is a network protocol for sensor networks . Sensor networks consist of tiny, wirelessly communicating computers (sensor nodes) that are deployed in large numbers in an area, network independently and monitor their surroundings in group work with sensors until their energy reserves are exhausted. As a special form of the ad-hoc network , they have completely different requirements for a network protocol than, for example, the Internet and therefore require network protocols specially developed for them. Sensor Media Access Control defines in detail how the nodes of a sensor network exchange data, controls access to the common communication medium of the network as Media Access Control (MAC), regulates the structure of the network topology and offers a method for synchronization .
Although today of purely academic interest and with no practical application, S-MAC was a significant step in sensor network research and inspired numerous subsequent network protocols. It was presented in 2001 by Wei Ye, John Heidemann and Deborah Estrin from the University of Southern California and is particularly aimed at conserving the scarce, non-rechargeable energy supplies of the sensor nodes. The development was financially supported by the US military agency Defense Advanced Research Projects Agency as part of the Sensor Information Technology (SenSIT) project.
classification
When S-MAC was introduced in 2001, there was already a broad spectrum of network protocols for digital wireless communication through mobile telephony and the then emerging technology of wireless local area networks (WLAN) for networking portable computers. Multiplex processes made it possible to operate large numbers of cell phones at the same time, newer devices established short-range wireless connections with Bluetooth, and in the computer sector IEEE 802.11g was the second expansion of the industry standard for WLANs.
Research into mobile ad-hoc networks, however, had not yet progressed so far that one could speak of predominant approaches or even foreseeable standards. The proposals made up to then modified the existing protocols more or less successfully. The Power Aware Multi-Access Protocol with Signaling Ad Hoc Networks (PAMAS) variation of the Multiple Access with Collision Avoidance protocol increasingly addressed the needs of mobile devices with limited energy reserves, but required an additional communication channel to prevent simultaneous media access. Time Division Multiple Access (TDMA) worked flawlessly in its raw form with Bluetooth with a maximum of eight end devices, but could not be easily transferred to large sensor networks. An unnamed TDMA variation by Sohrabi and Pottie made insufficient use of the available airtime and thereby drastically reduced data throughput . Piconet and various modifications of Carrier Sense Multiple Access (CSMA) showed promise, but were still immature.
S-MAC was suggested by PAMAS and IEEE 802.11 . It combines the two technologies Carrier Sense Multiple Access and Time Division Multiple Access in order to specifically address the special requirements of the sensor networks. In doing so, S-MAC takes into account the fact that, in contrast to the assumptions of PAMAS, there is usually only one communication channel available and breaks with the principle of fairness in computer networks . In normal computer networks, different users and processes compete for the common medium, so that fair media access regulation makes sense. In sensor networks, however, there is only one single, network-wide application that to a certain extent competes with itself, and its overall goal has priority over fairness towards individual network nodes.
Sensor networks have so far been a research area with no practical applications, so that S-MAC is also of exclusively academic interest. The main strength of the design lay in creating a starting basis for a whole range of similarly stored additional protocols: T-MAC , WiseMAC , DSMAC , B-MAC , μ-MAC , M-MAC and Z-MAC were not only based on names on their common pioneer. Among other things, S-MAC was implemented with TinyOS , an operating system specially designed for sensor nodes, and as the "ancestor" of the specialized sensor network protocols it is often used for comparisons.
Protocol structure
The protocol is based on the rendezvous technique , which is also used in other areas of computer science. A network node that wants to send a message first sends a send request (RTS, request to send ) to the addressee. If they are ready to receive, the person addressed answers with a confirmation of transmission (CTS, clear to send ). The sender then sends the actual data packet, which the receiver acknowledges with an acknowledgment of receipt ( ACK , acknowledgment ). This data exchange agreement assures the sender that the addressee is really listening and enables the recipient to refuse a data transmission without having to do anything, because he is currently communicating otherwise.
The rest of the protocol is divided into three areas of responsibility: daily routine, communication disruptions, and messaging.
daily routine
Sensor nodes can switch to a standby mode in which all components except for the clock are switched off. In this “sleep phase” their energy consumption is minimal, but they are inactive and in particular cannot receive any messages. S-MAC specifies a fixed "daily routine" for the sensor nodes with regular wake-and-sleep cycles. As many nodes as possible should follow the same daily routine in order to ensure secure communication during the shared waking time.
The daily routines are coordinated via synchronization impulses (SYNCs), short messages with the analogous content: “It is now ... clock. I'll go to sleep in x seconds. ”After switching on, each sensor node waits for a SYNC message from another node. If it does not receive one, it sends one itself after a random period of time. Whoever receives a SYNC adapts to the daily routine specified in it. If it was the first SYNC received, it determines the time difference d between its current time and the time of the message and sends a new SYNC with the content: “It is now ... clock. I'll go to sleep in x - d seconds. ”Each node remembers the daily routines of its neighboring nodes in order to establish contact with its own daily routine if necessary.
Since several synchronization impulses can start at different ends of the network at the same time or at different times, with this method the network can be broken down into networks that follow different day and night rhythms. At the boundaries of these groups there are sensor nodes that follow the daily routines of all neighboring networks. Although this ensures communication between the node groups, since the border nodes have more waking times than internal nodes, they consume more energy and fail earlier.
Even electronic watches are never exactly the same. It is therefore necessary to use renewed synchronization pulses to prevent the times from gradually "drifting apart" . For this purpose, S-MAC provides for the waking phase to be divided into two sub-phases: the shorter first is intended for sending and receiving synchronization messages, the longer second for the actual data exchange.
Frequently renewed synchronization pulses also make it possible to add new nodes to the network during ongoing operation. You receive a SYNC from the existing network and adapt to the daily routine.
Communication problems
Communication problems in computer networks lead to data loss and waste of energy and must therefore be avoided. S-MAC's rendezvous scheme naturally reduces the problem of hidden stations ; further components of the protocol to avoid other disturbances are shown below.
Data collisions occur when two nodes transmit simultaneously over the common communication medium: the signals overlap and both messages become unusable. To avoid collisions, S-MAC builds on physical and virtual carrier testing (Carrier Sense) , i.e. the ability to exclude the use of the communication medium before sending. The physical carrier test refers to the actual short listening to the medium in order to rule out any other use, while the virtual carrier test refers to the anticipation of other communication based on transmission requests announced in advance.
The virtual vehicle test also avoids overhearing . With overhearing, a node wastes energy because it is listening to data exchanges that are not intended for it and with which it cannot do anything. The virtual carrier check is implemented in that the length of the message to be sent is communicated at the same time in each send request. Listening nodes can conclude from this how long the communication medium will be occupied after the start of data transmission and can go to sleep for this time. This is using a network allocation vector (NAV, network allocation vector implemented). The NAV is set to the value communicated in the send request and gradually counted down using a timer. If the NAV reaches the value 0, the medium is free according to the virtual Carrier Sense and the node is woken up. After reassurance with the physical carrier sense, the node can then initiate its own data transmission.
Messaging
In computer networks, it is common to split large contiguous data units into small packets or fragments . If an error occurs when transmitting a large data unit, all the data must be sent again; if a transmission error occurs in a packet, it is sufficient to repeat the individual data fragment. This approach is also followed with S-MAC. Usually, the rendezvous scheme would now be applied at the packet level, that is, each individual packet would be confirmed with a complete set of RTS, CTS and ACK rendezvous messages. This unnecessary overhead is reduced in S-MAC by sending all related packets one after the other in one piece, so that only one RTS and CTS are required per complete data unit.
The receiver confirms every packet received with an ACK, and if this is not done, the sender repeats the last fragment. Here, too, one could limit oneself to a single ACK at the end of the overall message, but the developers have deliberately decided against it in order to build up a reinsurance against the communication disruption Hidden Station : The continuous sending of the ACKs prevents nodes in the recipient's vicinity from communicating to start, even if they have not heard of the RTS-CTS exchange of the communicating. For a further improvement, S-MAC recommends specifying the expected remaining duration of the data exchange in every packet and ACK; Newly added or "incorrectly" awakened nodes can go to sleep again.
Efficiency
Since S-MAC was one of the first comprehensive network protocols for sensor networks, the developers did not compare their method with other sensor network protocols, but with the industry standard IEEE 802.11 for wireless network communication. Since sensor networks form ad-hoc networks, 802.11 in its ad-hoc mode could also be used as a protocol for sensor networks. In fact, 802.11 was created for completely different requirements and, for example, does not address the issue of energy efficiency in any way. Accordingly, S-MAC is significantly more efficient than 802.11 in sensor networks.
S-MAC has two major weaknesses. On the one hand, this is the emergence of multiply loaded border nodes, already shown above, when the network is broken down into several networks by the synchronization pulses. Since the border nodes follow several daily routines at the same time, i.e. have more waking hours than other nodes, they consume more energy and fail earlier. The border nodes are particularly important, since without them the entire network breaks down into several individual networks.
On the other hand, it can happen that recipients go to sleep before the sender can contact them with a send request. If communication is currently in progress, another sensor node that wants to send data cannot make a send request. The addressee of the data is therefore not aware of the existing transmission request and may go to sleep before the sender comes to send the transmission request. This is where the S-MAC successor Timeout Media Access Control (T-MAC) comes in to achieve an increase in efficiency.
It was also criticized that S-MAC goes well beyond the goal of a MAC protocol by simultaneously implementing network initialization, synchronization, data fragmentation and hidden station treatment. According to the developers at Berkeley Media Access Control (B-MAC), these tasks should be outsourced to their own configurable modules.
Sensor network protocols developed later specifically addressed the weaknesses of S-MACs and, as expected, proved to be more efficient. Even the direct successor T-MAC was able to improve the energy yield compared to S-MAC through minor modifications. B-MAC outperformed S-MAC in terms of data throughput and energy efficiency. There are no large-scale comparisons of sensor network protocols; Some researchers doubt the usefulness of such studies and assume that there will never be a protocol for sensor networks , but always a selection of protocols that are differently well suited for different scenarios.
Web links
- Sensor-MAC (S-MAC): Medium Access Control for Wireless Sensor Networks - project page of the S-MAC developers
swell
Main source:
- W. Ye, J. Heidemann, D. Estrin: An Energy-Efficient MAC Protocol for Wireless Sensor Networks. (PDF; 183 kB) In: Proceedings of the 21st International Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM 2002) , New York, USA, June 2002. First published as technical report ISI-TR-543, USC / ISI, September 2001.
Individual evidence:
- ↑ a b s. Main source
- ^ W. Ye, University of Southern California: SCADDS: Scalable Coordination Architectures for Deeply Distributed Systems. Evaluated on March 23, 2007.
- ^ S. Singh, CS Raghavendra: PAMAS: Power aware multi-access protocol with signaling for ad hoc networks. In: ACM Computer Communication Review , 28 (3), pp. 5-26, July 1998
- ^ K. Sohrabi, GJ Pottie: Performance of a novel self-organization protocol for wireless ad hoc sensor networks. In: Proceedings of the IEEE 50th Vehicular Technology Conference 1999 , Vol. 2, pp. 1222-1226.
- ↑ J. Polastre, J. Hill, D. Culler: Versatile Low Power Media Access for Wireless Sensor Networks. ( Memento of March 6, 2007 in the Internet Archive ) In: Proceedings of the Second ACM Conference on Embedded Networked Sensor Systems (SenSys'04) , November 3 to 5, 2004.
- ↑ T. van Dam, K. Langendoen: An Adaptive Energy-Efficient MAC Protocol for Wireless Sensor Networks ( Memento of March 31, 2010 in the Internet Archive ) (PDF; 284 kB) In: Proceedings of the First ACM Conference on Embedded Networked Sensor Systems (SenSys) , Los Angeles, California, November 2003.
- ^ I. Rhee, A. Warrier, M. Aia, J. Min: Z-MAC: A Hybrid MAC for Wireless Sensor Networks. ( February 5, 2007 memento on the Internet Archive ) Technical Report, Department of Computer Science, North Carolina State University, April 2005.
- ↑ I. Demirkol, C. Ersoy, F. Alagöz: MAC Protocols for Wireless Sensor Networks: A Survey. (PDF; 244 kB) In: IEEE Communications. 44 (4), pp. 115-121. April 2006
