Professional Documents
Culture Documents
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com Volume 2, Issue 4, July August 2013 ISSN 2278-6856
A Learning automata- Based Protocol for solving congestion problem in Wireless Sensor Network
Raheleh Hashemzehi
Department of Computer, Sirjan Branch, Islamic Azad University, Sirjan, Iran.
Abstract
The phenomenon of congestion arises in a network when the senders rate of injecting the packets into the network is greater than the retrieving rate of the receiver. Congestion in a network can be determined by monitoring the node buffer occupancy and the wireless link load. Congestion leads to wastage of energy and minimizes the lifetime of sensor nodes. This paper suggests an approach for solving congestion problem. In the proposed model an algorithm based on learning automata each node has an automata that selects an action and adjusts its rate based on the responses receives from environment. One of the important features of this approach is learning from the past that is the reason it improves itself as time progresses.
importance of each sensor node. Nodes are assigned a priority based on the function they perform and its location. Nodes near the sink have a higher priority. The congestion is detected based on the ratio of sending rate to the packet arrival rate. If the sending rate is lower, it implies that congestion has occurred. The congestion information is piggybacked in data packet header along with the priority index. Nodes adjust their sending rate depending on the congestion at the node itself. PCCP tries to reduce packet loss in congestion state while achieving the weighted fairness transmission for singlepath and multipath routing.
Congestion Avoidance and Detection (CODA) CODA [5] is energy efficient congestion control mechanism designed for WSNs. CODA, detects the congestion by observing the buffer size of sensor nodes and the load of the wireless channel. If these two characteristic exceed from a pre-defined threshold, a sensor node informs its neighbour to decrease the transmission rate. Before transmitting a packet, a sensor node divides the channel to fixed periods. If it found the channel busier than pre-defined times, it adjusts a control bit to inform the base station of the congestion. CCF CCF [6]detects congestion based on packet service time at MAC layer and control congestion based on hop-by-hop manner with simple fairness. CCF uses packets service time to deduce the available service rate and detect the congestion in each intermediate node. When the congestion is experienced, it informs the downstream nodes to reduce their data transmission rate and vice versa. SenTCP SenTCP [7] is an open-loop hop-by-hop congestion control protocol with It uses average local packet service time and average local packet inter-arrival time to calculate local congestion degree .In SenTCP, each intermediate sensor node will send feedback to previous hop. This feedback contain local congestion degree and the buffer length and inform neighbor nodes to adjust rate .
1. Introduction
Wireless sensor network (WSN) plays an important role in many application areas like in military surveillance, health care etc. A wireless sensor network (WSN) [1] is a network made of hundreds or thousands of sensor nodes to cooperatively monitor different condition, such as temperature, sound, vibration, pressure and motion at different locations which is the common task of sensor node is to collect the information from the scene of event and send the data to a sink node. Many wireless sensor network [2] applications require that the readings or observations collected by sensors be stored at some central location. Congestion can occur while collecting the data and sending it towards the central location over the wireless sensor network. The phenomenon of congestion arises in a network when the senders rate of injecting the packets into the network is greater than the retrieving rate of the receiver. Congestion in a network can be determined by monitoring the node buffer occupancy and the wireless link load. 2. Review of the related works We review various congestion control approaches in the section below. Priority Based Congestion Control Protocol (PCCP) PCCP[3] [4] is a congestion control mechanism based on node priority index that is introduced to reflect the Volume 2, Issue 4 July August 2013
Page 396
Page 397
Figure 2. Network topology 5.1 Source traffic priority( SP (i)) Source traffic priority node i, expressed preference for the local traffic source node i. 5.2 Transit traffic priority (TP (i)) Transit priority traffic in sensor node i, for expression, the amount of traffic passing through node i is used. TP (i) is the sum of the first child nodes of origin. And all the children of node i depends on the source traffic priority. 5.3 Global Priority (GP (i)) Global priorities concerning the total traffic per node, equal to the sum of the priority source traffic and transit traffic . Equation( 4) GP(i)= SP(i)+ TP(i) 5.4 Rate adjustment of the proposed method Learning automata in each node acts as a decision maker, And trying to make the best of a set of activities that may be imposed on each node to the changing environment. In other words, a learning automata-based approach to learning , to identify an action / operation optimization, the set of actions is allowed. It is applied in a random environment, And the environment by producing an output action is applied, will respond. Environment is a triple E = {a, b, c}, The a ={reduced rate, increased rate} Action learning automata, B={Unfavorable,favorable}Set of environmental responses, c={c1,c2,.,c3}total probabilities fine, and ci are likely to receive an unfavorable response from the environment when i attempt to apply the environment, The environment response is binary. So (0,1) P model is chosen environment. In the beginning, the probability of choosing each equal action is 1/2. According to the above procedure to reduce the rate of convergences come about as follows: Equation 5 r) = New rates(r Where GP (pi, o) is Global priorities parent, and GP (i) is child node priorities. And the ratio of these two priorities than the priority of the child node to Global Priorities of the parent, In fact, its share of the bandwidth. This way, we give priority actions. Since the absolute priority of the child, is smaller than total priority of parent, The result of more than a small fraction of r ' is also previous rate. Rate increases as the operation is as follows: Equation 6 r)= New rates(r Volume 2, Issue 4 July August 2013
Figure4.congestion degree 7. Conclusion We are dealing with a host of challenging issues in wireless sensor networks like: limited energy supply of a sensor, which is a threat for network life time, QOS needs and energy efficient routing. If we equip the sensors with some intelligent automata, our network earns the ability of adapting itself with the environment when it changes. In other words, the WSN uses the power of learning and becomes intelligent as time passes. In this paper, we have proposed a new congestion control approach for sensor networks, Congestion severely affects the performance of a wireless sensor network in two aspects: increased data loss and reduced lifetime. The results show that our protocol performs better in terms of extending network lifetime. Our proposed algorithm does not require the source nodes to be fed back by the intermediate nodes to slow down. In our approach, we save this time, by making the intermediate nodes to take a proactive approach in controlling the current rate of flow of packets, thereby improving the overall performance of the network in the aspects of life time, energy consuming , congestion control and so on. References [1] XiangYangLi, Wireless Ad Hoc and Sensor Networks, Cambridge University Press, New York, 2008. [2] Akyildiz, I.F., Weilian, S., Sankarasubramaniam, Y., Cayirci, E.: A Survey on Sensor Network. IEEE Communication Mag. 40(8), 102114 (2002) [3] Wang, C., Sohraby, K., Lawrence, V., Li, B.: Priority Based Congestion Control in Wireless Sensor Networks. In: IEEE International Conference on Sensor Networks, Ubiquitous and Trustworthy Computing, Taiwan, pp. 22 31 (2006) Volume 2, Issue 4 July August 2013
Page 399