Communication protocols and techniques are often evaluated using simulation techniques. However, the use of formal modeling and analysis techniques for verification and evaluation in particular for Wireless Sensor Networks (WSN) becomes a necessity. In this paper we present a formal analysis of the backoff procedure integrated in the medium access control protocol named ECo-MAC designed for WSN. We describe this backoff procedure in terms of discrete time Markov chains (DTMCs) and evaluated using the well known probabilistic model checker PRISM. After checking the different invariants of the proposed model, we study the effect of contention window length (in number of time contention unit) on the acceptable number of simultaneous senders in a neighborhood of a given receiver. The obtained quantitative results confirm those provided by the simulation using OPNET tool and justify the validity of the adopted value for the time contention unit TCU.
Wireless sensor networks (WSN) is composed of many sensor nodes, scattered throughout a target area, that monitor some events witch depend of the application characteristics. Collected data by these sensor nodes has to be forwarded to a destination node called Base Station (BS) [1]. WSNs have been an active research topic during the past few years because they have been proposed for a large range of application and they have some constraints such as limited battery lifetime, reduced memory capacity, etc. Researchers have mainly conducted extensive studies on energy saving and other performance criteria using the simulation techniques and discrete event simulators.
Constraints induced by the wireless medium motivate the implementation of new communication protocols characterized by a probabilistic behavior. This characteristic cannot provide complete coverage of a model proposed using simulation environment. To overcome these limitations, model checker techniques are proposed [15]. These techniques can be used to provide an exhaustive analysis and an automatic verification of partial or abstract model. Therefore, model checker requires the use of abstracted or simplified model that must be true to the real system behavior. Several frameworks can facilitate modeling and checking process such as UPPAAL (Uppsala University and Aalborg University project) [17], SPIN/PROMELA (Simple PROMELA INterpreter/PROcess MEta LAnguage) [19], PRISM (PRobabilistIc Symbolic Model checker) [18], etc.
In wireless communication, several medium access control protocols use a backoff procedure in retransmission phase to reduce the likelihood of reappearance of a collision. Different procedures are described and proposed such as the Binary Exponential Backoff (BEB) used in the international standard IEEE 802.11 [3], the normal distribution introduced by Gobriel et al. in [2], the Impatient Backoff Algorithm (IBA) studied in [4], the Multiplicative Increase Linear Decrease (MILD) [5], the Double Increase Double Decrease (DIDD) [6], etc. In WSN we had proposed a new medium access control protocol ECo-MAC [7] with a new backoff procedure. We had studied the effectiveness of the proposed procedure using OPNET [16] simulation.
Different works have studied a BEB backoff procedure by the simulation technique [8], by the analytic approach [9] and using the UPPAAL and PRISM model checkers [10]. In this paper, we consider automatic verification and quantitative evaluation of the backoff procedure of ECo-MAC protocol implemented in OPNET simulator. For probabilistic modeling and analysis, we have used the probabilistic model checker PRISM framework. This tool has proven to be successful in a wide range of case studies [10], [11], [12], [13], [14].
The paper is organized as follows. We begin by presenting the proposed backoff mechanism with the ECo-MAC protocol in section 2. Then, we describe the proposed probabilistic model of this procedure in section 3 using PRISM framework. Finally, we analyze the proposed model with PRISM tool in section 4, 5 and 6. We terminate the paper by a conclusion in the section 7.
As described in [7], ECo-MAC is a media access control protocol for wireless sensor networks. This protocol is hybrid. It takes advantage of three access techniques CSMA, TDMA (Time Division Medium Access), and multi-channel protocols. Differently of classical TDMA protocols that manipulate two level of time division (frame and time slot), ECo-MAC uses only one level of division into different time slots (TS). All time slots have a constant length that is depend on radio TX/RX characteristics and different protocol parameters. Considering that the majority of WSN applications are characterized by a low traffic, authors of ECo-MAC decide that at a beginning of each TS, all nodes having data to transmit send immediately an RTS without listening to the channel. This makes all nodes, except the destination addressed in the RTS to switch to the SLEEP mode for the remaining time in the currents TS. Two situations may occur as an RTS overcome. (1-success) The source receives the expected CTS from the destination node and then the communication takes place within a sub-band frequency randomly selected by the source node and transmitted in the RTS packet. (2-fail) The source node receives a JAM packet or the receiving response timer expires without receiving a valid CTS frame, meaning that a collision occurs. To avoid collision in the retransmission of current data packet, node executes a backoff procedure to reduce a possibility of collision in the retransmission phase in current TS.
Sender nodes that fail the free access in the first sending phase, must execute a backoff procedure in a second sending phase in the current TS. In contention phase, each deferring node chooses backoff duration composed of an integer random number (rbc) of time contention units (T CU ). Value of rbc is drawn from a uniform distribution uRand over the interval [
This content is AI-processed based on open access ArXiv data.
