Node Level Congestion In Wireless Sensor Networks Computer Science

Essay add: 30-03-2016, 17:49   /   Views: 2

Abstract-The Application specific wireless sensor network differs basically from the general data network. It focuses on tight communication but restricted in storage, lifetime, power and energy. The WSNs consists of unbelievable network load and it leads to energy wastage and packet loss. Many of the existing concepts are developed for link level congestion control. The Rate optimization technique for node level congestion will assist to control the traffic at node level. Except source and sink node the remaining nodes may participate in forwarding the packets towards the communication direction. The rate based adjustment technique is applied to avoid packet dropping in order to save the network resources. We are proposing this scheme to avoid the buffer overflow and it is not taking too much energy consumption in the communication. This scheme will assist to improve the throughput, efficiency and resource saving. Node level congestion control is effectively needed for WSN, because the node deployment can be anywhere. We are Introducing this scheme using the network simulators extended tool called mannasim.

Keywords-BufferOverflow;CongestionControl;Node deployment; sink.


Wireless sensor Networks have been used tremendously in target monitoring, location tracking and battlefield etc. Generally the sensor nodes are restricted in storage, lifetime, power, computation capability and in energy too. It can be deployed anywhere, where it might not be possible to use traditional networks. The formation of the network is done without human intervention. Similarly addition and removal of nodes is done without human help. We can operate these nodes in uninterrupted environment in order to accomplish the task. The sensor nodes are very small in nature, they consists of small processing unit, which is having the capability of limited computational power [9].

Generally they are battery powered and energy constrained. The sink node i.e. target node is most powerful which is used as gateway to the wired network and it is doing data collecting and processing. The nodes deployed area is called sensing field, where each node has its own range. The lifetime of the sensor node is normally limited, before its energy exhausted, it should be used effectively. In many of the applications, less cost sensors are used and placed in the vast region [1].

The routing technique used is challengeable one. The ad-hoc routing protocols are not applicable [4], because here nodes

are numerous and power constrained. The topology also random topology since it is changing its structure frequently. During the communication flow, traffic is unavoidable one. The traffic in WSNs causes the congestion in the network. In previous research fairness bandwidth allocation is not well addressed. But it is very difficult to allocate the rate for each flow in communication. Congestion in WSNs has negative impacts on network performance and application objective, i.e., packet loss, increased packet delay, wasted node energy and severe reliability degradation.

Figure 1Simple wireless sensor network

However, some characteristics of WSNs, such as constrained resources, interference of paths and the lack of centralized coordination, make the congestion problem in WSNs more than traditional networks. In addition, to let the sink successfully receive the data from different sensors, we need to consider the fairness issue among the source nodes [6]. The main source for network congestion in WSNs can be Channel Contention and Interference. Another one cause can be a Packet Collisions. Increasing network contention causes an increase in packet collisions in the wireless medium. After

Several unsuccessful retransmissions, these packets are

dropped at the sender node. In order to control the congestion initially congestion detection is essential.

In WSNs, congestion can be detected by two main factors such as buffer overflow and link collisions. To achieve the congestion less communication in sensor networks, we have to do rate based adjustment in the communication path. We are introducing all the intermediate nodes under two classifications. The nodes which are surrounding the source and sink are mainly categorized. The nodes which are surrounding the source node are called source surrounding nodes and the nodes which all are very near to sink is called sink surrounding nodes. These intermediate nodes are going to control the fair data rate in the communication.

The wireless sensor network not yet reached high, since it is having problem of flow control, rate control, congestion control, medium access control, queue management, power control and topology control. It is very difficult to offer complete solution for solving all these issues. The event driven wireless sensor network is generating numerous packets when particular event is occurred. After even detection the data is propagated towards the base station. Once the flow of data is not regular, then we can decide that congestion is occurred.

The remaining part of this paper is continued with related works, models and problems, proposed scheme and evaluation. Finally we conclude this paper.

related works

In the literature, survey works have been conducted on congestion detection, congestion control and communication models in WSNs.

When congestion occurs, the source and destination communicating with overflow of packets. Generally the destination can keep or drop the packets. But a congested node can drop the packet after overflow of the buffer. There are two options to drop the packet. The destination can drop the arriving packet and maintains the old packets. Otherwise it will drop the old packets and keeps the arriving packets. Anyway old and new packets are replaced with each other. Recently many congestion controlling techniques have been introduced.

The congestion can be detected easily by checking buffer occupancy and channel occupancy. Mitigating Congestion in Wireless Sensor Networks (Bret Hull in 2004) used hop by hop congestion control and network level congestion control. But the fairness of the network is under load and performance comparison is left for future work. On The Interdependence of Congestion and Contention In Wireless Sensor Networks (Mehmet C. Vuran Vehbi C in 2004) investigated widely about congestion in wireless sensor networks and their experiment result says interaction between the nodes directly affects the performance. But the buffer size is recommended to be small; it is not useful for loading more data during the communication. Recently network simulators extended tool mannasim is widely used for wireless sensor networks simulation. But the results are not yet produced for wireless sensor networks congestion control using this mannasim and its script. It provides the various sensor node parameters and configurations for energy, throughput and dropping ratio. It is the developing environment tool for wireless sensor network. Apart from this, the link level congestion scheme only widely discussed in recent papers. But node level congestion controlling scheme is new idea in wireless sensor network using mannasim tool.

models and problems in communicationA.NODES FORMATION

Wireless sensor nodes are deployed randomly and it consists of large number of nodes naturally. The nodes share the wireless medium to communicate with neighbor nodes. If the nodes are placed within the node's range, then it can communicate easily with all others. But generally sink may be placed anywhere in the network. In order to contact the sink it needs to communicate through the intermediate nodes. The intermediate nodes become busy, when they are trying to forward the packets more than its capacity [2].

If it receives more data than it can forward then congestion occurs in the network. So the characteristics depend on the generating rate of the data's from the source node. If the generating rate is low then the interference makes less severe.


The base station or the sink is not always near to the source nodes. It is usually a gateway to the outside network. The nodes are facing the difficulties to find the optimal path towards the destination node. It passes the packet through the intermediate nodes. The mediating nodes are wasting their energy in order to find the sink node. If the target is too far from the source means some amount of energy is spent for finding the path.


Normally the buffer is used for storing the information temporarily. Therefore the node's buffer size is main consideration in the network. If the size too small, more packets are sent out of the network. Similarly if the size is too large more replications will occur. If some of the packets are dropped, we can conclude buffer size may be small. If the link is not reliable, then we can realize that buffer size too large. Ultimate aim of the buffer is to hold the excess packets, even though if it is dropping the packets which indicates congested communication.

The congestion control is followed to get rid of the congestion. But the congestion avoidance is used for preventing congestion [5]. When the network load increases the packet loss possible also increased. To control the sending rate of the data packets from the source and intermediate nodes, we have to adjust the generating rate of packets. The efficient rate allocation may assist to control the congestion in the network [3].


In networking concept, network congestion occurs when a link or node is carrying so much data so that its quality of service is degraded. Typical effects include queuing delay, packet loss or the blocking of new connections. Consequence of these two leads to reduced network throughput, or to an actual reduction in network throughput [7].

Figure 2 Logic Scheme for Buffer Overflow and Link Collision

There are two reasons mainly affecting the network throughput that leads to congestion in the communication. Those factors are buffer overflow and link collision. In order to control the congestion generally two techniques are followed in the network concepts. Resource management and traffic control are the techniques handled to control the congestion.

Buffer overflow-when a sensor node try to receive more data than it can forward, it leads to buffer overflow. Due to overflow, some of the packets are dropped while forwarding. Normally intermediate nodes are doing this packet forwarding. Link collision-when multiple nodes are trying to seize the channel simultaneously, the link may get polluted state. The shared channel becomes busy and multiple nodes' link is getting collision. Our logic scheme shows the buffer overflow and link collision situations in wireless sensor networks. The solution for controlling the congestion is described by two logics called resource management and traffic control. The resource management tells to increase the resources which are used in the communication. The packet delivery ratio shows the best performance of the network.

Packet delivery =

Number of packets received

(Total number of packets transmitted by previous node)

Resource management is difficult in WSNs, while deployed in the random places. Another logic traffic control is using the rate adjustment technique. It is adjusting the traffic rate at source or intermediate nodes. This implies the data rate adjustment in the communication. Traffic control saves the network resources effectively and provides efficient communication in the network. The traffic control further divided into classifications such as end to end control and hop by hop control. The end to end control indicated the exact rate adjustment in the source nodes. On the other hand, hop by hop control adjusts the packet forwarding rate at the intermediate nodes.

RATE OPTIMIZATION SCHEME DESCRIPTIONCongestion Control at Node Level Congestion:

For control the node level congestion, the node is identified initially which has met the congestion. Due to congestion it drops some of the forwarded packets. Controlling logic is applied to that particular node to get eradicated from the traffic. The source surrounding nodes initially receives the packets from the source nodes. Source surrounding nodes are the nodes which are very close to source nodes. The source nodes may select any of the source surrounding nodes to forward the packet towards the base station. I.e. normally sink in the WSNs. Further these nodes forward the packets to the sink surrounding nodes. The sink surrounding nodes are the nodes which are very close to the sink nodes.

The scheme shown in figure3 is the logic used in this technique. The congested node drops some of the packets and it receives the warning message from the next nodes. Now the congested node realizes that it has dropped some of the packets. The message leads to consequence which provides warning message to the source node.

The warning message is sent when the packet is dropped or the nodes receive more than they can forward. In order to adjust or update the data rate the rate optimization is applied. Our aim is to control the buffer overflow. When the incoming rate is greater than its capacity then it leads to overflow. Now the source node adjusts its data generating rate in order to avoid the buffer overflow in the communication. The algorithm used for this scheme is described below. Each sensor node watches its buffer and notices the congestion bit.

Figure 3 Logic frame for node level congestion control

The buffer size is measured in terms of packets and data rate is measured by number of packets transmitted per unit time.

The algorithm illustrates that, every sensor node has individual buffer, and its buffer level is continuously listened. Introducing both minimum and maximum threshold values of the buffer occupancy called MinBo , MaxBo respectively. The present buffer occupancy is indicated by CBo and it is compared with extreme values. If the current level is exceeding the maximum extreme means then set the congestion notification bit in the packet header. So the

previous node will decrease the transmitting rate. Similarly if Present value goes beyond the minimum extreme then the

Previous node can update its transmitting rate.


Listen the buffer level continuously.

Set minimum & maximum extremes of the buffer occupancy, say MinBo , MaxBo

Find the present buffer level and let it be CBo.

If CBo exceeds the MaxBo , then set congestion notification Bit.

Decrease the data rate of the previous node.

If CBo goes beyond MinBo level, then reset the congestion notification bit.

Increase data rate of the previous node.

Buffer Representation:

Figure 4 indicates that sensor nodes are receiving the packets from the source nodes, and it is stored in its local buffer. The buffer is having two portions normally, one is for storing the incoming data and another is for loading the data for transmission. Every node is processing the incoming data and the processed data is put into the loading portion. If the incoming rate is slow, the data is well processed. So the flow will be normal one. If the incoming rate is too fast, the node fails to process the data properly and dropping some of the packets.

Figure 4 Sensor node's Buffer

In order to keep the flow in regular the rate should be maintained in good manner.

Data flow in sensor nodes:

The flow diagram explains the flow of the data followed in each sensor node. Every node checking its buffer occupancy level with extreme values as explained in algorithm. The decision is made based on the logic result. If the result is 'adjust' then the transmission data rate of the previous node is decreased. If the result comes with 'update', then the data rate is increased since the node is having the capability to hold the extra amount of data.

Figure 5 Data flow in each Sensor node

Performance evaluation

The moderate simulator ns2.29 with extended tool mannasim, the sensor networks properties is defined. Routing protocol is selected by users wish, but the MAC layer is selected and number of nodes selected is varied. The parameters used for the simulation is shown below. The queue occupancy plays a vital role in the simulation environment. We should define the traffic flow through them. The link definition must include the way to handle the queue overflow. In case buffer over capacity is exceeding the capacity various options are handled such as Drop tail motion, RED, FQ, SFQ, CBQ and DRR. In drop tail motion the exceeding size is considered as last arriving packets and they are dropped. For wireless sensor network we are defining UDP connection with CBR application. The RED is using the idea that the packet should not wait till the buffer is full. But it is detecting the congestion before the buffer overflows. We evaluate the congestion occurrence rate for various conditions using network simulator ns2 version 2.29 to conduct the simulations. The default simulation parameters are described below. We implement the simulation environment for comparison purpose.

The network performance can be determined by few terms such as channel busy time, channel utilization level, efficiency, fairness and imbalance. The amount of the time the channel is allocated for packet transmission and reception is called channel busy time. Similarly channel is sometimes being idle during communication. The unit of time which makes delay to transmit a packet is called channel access delay time.

The channel or medium utilization level can be defined as average rate of reliable packets delivered through the channel. The MAC layer utilization level can be determined by noticing whether the medium is busy or idle. The binary values are used for indicating MAC layer utilization level. 1, 0 are used for indicating channel is now busy or idle respectively. The main factor deciding buffer overflow is interface queue length. The extra amount of packet are buffered in the interface queue only, the packets are dropped when the queue length is limited in the network. The main terms that are to be calculated to determine the network performance are efficiency, fairness and imbalance. The efficiency of the communication is calculated by number hops the successful packets travelled to the total number of packets placed (dropped and retransmitted also included) in the network

1. Efficiency = Number hops the successful packets travelled

Total number of packets transmitted.

It can be represented as,

Where 'j' is the hops varied from 0 to n and 'S' is the successful packets and 'A' is all the packets transmitted from node 1to n.

Fairness of the specific network can be calculated as,

2. Fairness = (Total number of packets sent from all nodes)

Total nodes * (Packet sent from each node)2

It can be represented as,

Where 'i' varied from 0 to n and it indicates the individual sensor node and 'p' is the packet sent.

Imbalance is the next factor to be calculated for determining the performance. Per-node the imbalance is determined; it is ratio between the node received packets to total number of successfully delivered packets.

The imbalance can be represented as,

Imbalance =

The node received packets

(Total number of successful delivers from the node.)

The ratio between received packets and successful delivers tells the imbalance of the current node.

We implement this scheme for comparison purpose in the scenario. As we know that node level congestion will decrease the efficiency and increases the dropping packets. When comparing with our proposed scheme the qualities are improved in the network. We first compare with default parameters and controlled buffer size. Compare the schemes in terms of dropping and throughput. Finally we simulate the scenario for 15 sensor nodes and run the simulation for 60 seconds for the comparison. The default network parameters are listed below.

Packet dropping comparison:

The simulation results shows the dropping ratio of the 'Without congestion control' and congestion control

Scheme. The figure shows the packet dropping value in three dimensional views. Our scheme gives reasonable improvement in throughput and reduced packet loss. In congestion control scheme the buffer overflow is mainly avoided in order to reduce the packet loss. Proposed scheme allocates the fair rate to avoid the congestion and improves the communication. Ultimate aim is to save the network resources and less energy consumption. When the data generating is higher in the source nodes the overflow is possible. After controlling the overflow, the generating rate is updated or adjusted according to the nodes capacity. The congestion model is struggling with more packet dropping as shown below in figure. 15 nodes show their ratio of packet loss during the communication.


simulation parameters



Number of sink nodes


Number of common nodes


Transmission range

250 M

Sensing range

550 M

Channel Bandwidth

1 Mbps

MAC protocol


Routing protocol


Simulation area


Network interface Wireless

Packet lost in proposed scheme without congestion control

It leads to the large number of dropping packets during the simulation for the specific parameters. As we can see, our congestion control scheme leads to reasonable fair rate with normal results. Figure 4 show the result of our scheme and figure 5 shows the result of normal scheme. Now we can decide that packet loss is increased when congestion is not controlled.

Efficiency of the communication:

The simulation proves that our scheme achieves higher throughput than without congestion control scheme. The throughput is determined using the sending and receiving packets comparison. When the network load is offered the source surrounding nodes forward it to the near nodes. While increasing the number of hops in the network, the efficiency is increased.

The efficiency is determined by dropped packets as well is retransmitted packets. When the traffic load increases efficiency is decreased.

Figure 6 Graph of efficiency to number of hops

Fairness:It is the metric to determine the communication link is good or bad. Therefore the ratio is taken for the packets sent by all the nodes to individual nodes transmitted packets. The consistent fairness shows that link is good as well as communication is reliable.

Figure 7 Graph of Fairness to Number of nodes

Generally the fairness is indicated in terms of percentage, our proposed scheme achieves reasonable fairness. The fairness shown in above graph, it is mainly affected when numbers of nodes are increased. It was slightly damaged when the number of node is increased in the scenario.


We have proposed this scheme to control the node level congestion in the wireless sensor networks. This scheme achieves reasonable fair data rate in the communication. The simulation scheme shows better performance in terms of efficiency, reliability and resource saving. Particularly for improving throughput it will assist while network load is increased. This scheme becomes reliable for avoiding packet loss during the congested node placed in the communication. In order to make the fair rate among the nodes, this scheme is applicable one. It aids to control congestion at node level particularly at intermediate nodes.

Article name: Node Level Congestion In Wireless Sensor Networks Computer Science essay, research paper, dissertation