Performance analysis and evaluation of REFIACC using queuing networks
Introduction
A Wireless Sensor Network is a set of hundreds or thousands tiny devices scattered in order to monitor a specific area and forward the happening events to a special unit, named sink or base-station. After the event happening or in a dense deployment of a periodic monitoring scenario, the huge information sent by the nodes without appropriate control leads to buffer overflow and consequently congestion happening. In addition, the fact of being in the same wireless vicinity of each other, the nodes communication interfere at the sending moment which leads to packets collision and loss. In [1], the need of congestion control schemes was motivated and different works that treat congestion control in WSN field have been presented. This control is usually done at transport layer, in addition to the use of cross layer information, especially for WSN characteristics. The use of scheduling schemes to manage congestion seems to be a good idea, conditioned by the awareness about the different causes of congestion and packet loss. This loss of data leads to throughput degradation, which is the number of packets received correctly by the base-station in a period of time. The principle of communication scheduling is to organise the nodes communication as a number of slots where in each slot, a node or a group of nodes send their data to the parents. In [2], the basic idea of a scheduling scheme that aimed to avoid energy wastage due to buffer overflow and links interference has been presented. While in [3], a global scheme to control congestion in wearable health monitoring systems that uses buffer length and channel state as congestion indication [4] has been presented.
In general, most of proposed transport layer works in the domain of WSNs based their evaluations on simulation results. Even this method has many advantages, but evaluating the performance of the proposed solution analytically brings more robustness and clarity. This argument motivated us to conduct modelling analysis to evaluate REFIACC scheduling scheme, that was already validated by intensive simulation [5]. In this study, REFIACC scheme has been modelled using stochastic approach based on Markov chains. The aim is to model the steady-state of system throughput and the queue length of different nodes. The sending process is modelled by a continuous-time Markov chain (CTMC) in the terms of queue length. The parameters of MAC and physical layers have been taken into account by modelling the loss due to wireless channels. The use of schedule-based scheme has as a goal to decrease losses due to collisions and so retransmissions, especially with bursty traffic which uses generally high rate sending. The throughput evaluation has also shown the delay resulted from REFIACC scheme use. Also, through the queue length it was shown the retransmission due to channel loss only and not due to collisions.
The rest of the paper is organised as follows. Section 2 describes some related work that interest on controlling and modelling congestion and reliability in Wireless Sensor Networks. In Section 3, the applications that need such a schedule method are presented. Section 4 highlights REFIACC behaviour. In Section 5, REFIACC modelling using queuing networks is performed. While in Section 6 the case of having a finite number of retransmissions is modelled in more details. Section 7 shows the numerical performance of the model on MATLAB. Finally, in Section 8 the paper is concluded and the work summarised.
Section snippets
Related work
With the incremental use and success of WSN, many works were interested in modelling their proposed systems according to the applications requirements. Congestion control and reliability assurance fields, being hot challenges topics, had taken the researchers attention, as the state-of-the-art works shown [1], [6], [7], [8], [9], [10], [11], [12], [13], [14].
In [6], TARA (Topology-Aware Resource Adaptation) proposes a schedule based on spatial interference graph and uses graph-colouring to
Targeted applications and existing shortcomings
We are interested in applications where each node has huge quantity of data to send through the multi-hop network towards the base-station. This can be the case in tracking objects using nodes’ camera, where each node sends this video or photo to the base-station. Also, continuous sensitive monitoring applications where each node sends small data but in a continuous manner in order to ensure data freshness, like temperature in nuclear stations or vibration in structures and bridges, are
Brief description of REFIACC scheduling and functioning
The aim of REFIACC is to construct a schedule to be used by the wireless sensors in order to forward sensed packets to the base-station. The advantage of REFIACC is to be aware about interferences and losses at the schedule construction moment. Actually, REFIACC could be considered as a transport layer based schedule that works in a cross layer like manner. The use of threshold for the schedule reconstruction is left for future works. Thus, REFIACC relies on any routing protocol that constructs
Modelling REFIACC behaviour as a network queuing system
In this section, REFIACC has been modelled as an open, single class queuing network with finite capacities. Each node is considered as a finite capacity queuing system. To simplify the model, a slot will be considered as the elementary sending period of one packet regardless of its success or not, and each node queue state is modelled. In the following, some hypotheses to highlight our consideration:
- •
Each node has a finite buffer queue where the packets are stored before transmission. The
Model description of finite retransmissions based model
In this section, the model is described by defining the stochastic processes and by building the associated Markov chains. Indeed, the proposed model is a queuing network with tandem queues with feedbacks like that given in Fig. 2.
Each node is considered as a finite capacity queuing system, where the finite buffer queue contains the packets to be transmitted. During each slot, one packet is sent by the concerned node to the corresponding parent regardless of its success or not. The success
Numerical results
In order to compare the conception of REFIACC schedule scheme and that of TARA (without resource control part) or FLUSH (without continuous interference probing) by Markov chain modelling, the results were computed numerically using MATLAB [22], [23]. The difference between the schemes and the proposed schedule is that of being aware about different capacities (FLUSH and TARA shortcoming) and interferences (TARA shortcoming). This behaviour is translated by giving an arrival rate higher than
Conclusion
WSN applications know more and more success and use either in civil or military applications. The throughput problem is important to handle in order to answer the applications requirements. Congestion and reliability control are for high importance to fulfil that exigency. In a previous study, a scheduling scheme, dubbed REFIACC, that tackles reliability and congestion problems, was proposed and validated by simulation. In this study, REFIACC scheduling scheme was evaluated analytically using
References (23)
- et al.
Interference-aware congestion control protocol for wireless sensor networks
Procedia Comput. Sci.
(2014) - et al.
REFIACC: reliable efficient fair and interference-aware congestion control protocol for wireless sensor networks
Comput. Commun.
(2016) - et al.
On reliability modeling in wireless sensor networks – a review
IJCSI Int. J. Comput. Sci. Issues
(2012) - et al.
Congestion control protocols in wireless sensor networks: a survey
IEEE Commun. Surv. Tutor.
(2014) - et al.
CCS_WHMS: a congestion control scheme for wearable health management system
J. Med. Syst.
(2015) - et al.
Congestion detection strategies in wireless sensor networks: a comparative study with testbed experiments
Procedia Comput. Sci.
(2014) - et al.
TARA: topology-aware resource adaptation to alleviate congestion in sensor networks
IEEE Trans. Parallel Distrib. Syst.
(2007) - et al.
Flush: a reliable bulk transport protocol for multihop wireless networks
Proceedings of the Fifth International Conference on Embedded Networked Sensor Systems, SenSys
(2007) - et al.
A capacity aware data transport protocol for wireless sensor network
Proceedings of the International Conference on Computational Science and Its Applications, ICCSA, Lecture Notes in Computer Science
(2009) - et al.
TDMA scheduling with maximum throughput and fair rate allocation in wireless sensor networks
Proceedings of the IEEE ICC 2013 – Ad-hoc and Sensor Networking Symposium
(2013)
Traffic aware scheduling algorithm for reliable low-power multi-hop IEEE 802.15.4e networks
Proceedings of the 2012 IEEE Twenty Third International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC)
Cited by (7)
A stochastic mobility model for traffic forecasting in urban environments
2022, Journal of Parallel and Distributed ComputingCitation Excerpt :Each lane of a road is modeled as a fixed-capacity queue and multiple servers each holding a vehicle for a specific service time representing the time needed to travel from the start intersection to the end intersection. The probability distribution used to model the arrival and departure of vehicles is similar to works in [20,24,43]. The vehicles arrive from the outside network to an intersection within the studied city section following a Poisson process.
Feasibility and modelling for convergence of optical-wireless network – A review
2017, AEU - International Journal of Electronics and CommunicationsCitation Excerpt :A new decentralised optimised Fi-Wi routeing algorithm has been proposed [73] to improve the aggregated maximum throughput. Recently, The WSN modelling based on queue theory is reported [74] and analytically evaluated for Reliable, Efficient, Fair and Interference Aware Congestion Control (REFIACC) in term of queue length and throughput. The summary of queue based modelling of wireless and optical communication network proposed the by various researchers are tabulated in Table 3.
Markov Chain Mobility Model for Multi-lane Highways
2022, Mobile Networks and ApplicationsA Stochastic Traffic Model for Congestion Detection in Multi-lane Highways
2021, Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICSTContinuous Time Markov Chain Traffic Model for Urban Environments
2020, Proceedings - IEEE Global Communications Conference, GLOBECOMPerformance analysis of non-exhaustive wireless sensor networks based on queueing theory
2020, International Journal of Communication Networks and Distributed Systems