Routing and Congestion Control in Wireless Sensor Networks using the Ant Colony Optimization algorithm and the Tabu search algorithm

doi:10.21203/rs.3.rs-2453157/v1

Wireless sensor networks (WSN) are a new generation of networks that consist of a enormous number of sensor nodes. These nodes are usually inexpensive and have the ability to communicate wirelessly, compute, and sensing. The main purpose of these networks is to collect information about the environment. These networks' function is that the nodes collect the information, and then this data is sent to the base station. Therefore, data transfer, routing, congestion control are important and challenging issues in this network type. Various protocols have been proposed in this field, considering energy utilization and increasing the network's lifetime are important problems. Our goal in this study is to consider the energy consumption and increase the network lifetime and some of the Quality of Service (QoS) parameters, such as increasing the number of alive nodes and throughput. In this paper, we use an Ant Colony Optimization (ACO) algorithm to try the routing send data from a source node to the Cluster Head (CH). Then the routing analysis performed with the Tabu search (TS) algorithm is to found the route that has congestion. If the path has more than a 70% threshold, the path putting up on the congestion route, and the ACO algorithm must find the new path to send data from source to CH. In this research, the proposed method's performance with the Adaptive Cuckoo Search Based Optimal Rate Adjustment (ACSRO) algorithm, Fuzzy based Cross-layer mechanism using Oppositional Artificial Bee Colony (FCOABC) algorithm, and the Flock-based Congestion Control (Flock-CC) algorithm in the same data set has been investigated in the NS3 simulator.

Recent advances in Wireless Sensor Network (WSN) and microelectronics have helped develop sensor nodes that are fast, independent, self-aware, and self-tuning [1]. These nodes are deployed across an area to sense the specific or unnatural event of the environment in the form of data such as humidity, motion, heat, smoke, pressure, etc [2]. These sensors can be networked and deployed in remote and inaccessible environments to establish stable communications[3]. The parameter measurement is sent to sink or other nodes through the network architecture. The sensor node of a small device with some essential tools such as the unit of measurement: to access the environment's data, the processor for data processing and storage for a transient purpose, the transmitter and receiver are radio frequencies to transmit the processed data[4]. WSN can be found in a wide range applications like monitor physical, environmental conditions, medical services, detection of air pollution, and numerous different applications where infrastructure is not fixed or readily available[5].

Congestion is an issue in packet-switched networks, in which the resources are more than the network capacity in one or more intermediate nodes (routers) and cause packet loss[6]. If the load provided to a network is not controlled, congestion collapse can occur[7]. The characteristics of WSN are affected by congestion, it is better to measure in certain conditions[8]. New congestion management protocols are aware of the nature of applications and can prompt more powerful solutions than classical strategies [9]. Congestion control is essential in resource allocation. Basic control congestion ensures that the source rate is adjusted to reduce or prevent congestion[10]. Better methods try relatively well to ensure that the sensor nodes can allocate resources in competition with other nodes and, during congestion, achieve shared bandwidth evenly[11]. One of the significant congestion control purposes in WSN is to give an ideal degree of reliability at the destination node [12]. In a way that energy efficiency and reliability ensure the successful delivery of data from the first node to the destination[13]. In many WSN algorithms, the focus is on avoiding or reducing congestion. The important purpose of providing congestion control is to limit energy utilization in the network and focus on WSN protocols' reliability [14] .

This paper describes selecting the best data transfer path from the node to a Cluster head (CH) and preventing data congestion in WSN. We have developed an advanced method for combining services by combining the Ant Colony Optimization (ACO) algorithm and the Tabu search (TS) algorithm. First, in this method, the best and closest path is selected by ACO algorithm, and then the pheromone updates the rout. After establishing the path, the TS algorithm starts working, finding the highly traded paths, and measuring the threshold. If the threshold is higher, the node and the path will be included in the tabu list, and the routing will be performed again. Briefly, the present article contributes to the following:

Increase the network lifetime and throughput by the proposed algorithm.
Load balancing between the nodes and reduce energy consumption.
Put high-transaction routes in the tabu list and replace the new route from the sources node to CH.

The remainder of this paper: Part 2 presents the related work. Part 3 provides the proposed algorithm. Then the experimental dataset and simulation outcomes are provided. Lastly, Part 5 presents the discussion and upcoming studies.

Numerous kinds of research about WSN congestion control have been performed. This part refers to certain papers that investigated about the congestion control to feature their fundamental points of interest and impediments. In addition, according to the importance of congestion contorol, especially when using a wide range of nodes, Quality of service (QoS) factor has been considered in analyzing the previous works as an important property. Notwithstanding the numerous protocols proposed to control congestion, this issue stays unresolved. Some of the significant investigations related to this research are listed below. Ordinarily, three workmanship can manage this problem:

2.1 Congestion Control Strategy

In WSN predominantly two reasons that bring about congestion first the packet appearance rate indulgences the packet service rate. This happens in the nodes sensor which are occurred in the nearby the sink as they normally convey more upstream traffic. Second dispute and intervention bit fault rate on a link consequence in congestion. Congestion straightforwardly effects on energy consumption and application QoS.

Various congestion control protocols are envisaged for high traffic on WSN. Several congestion detection, congestion notification, for rate adjustment mechanisms have been investigated. Portions of the protocols have been examined beneath:

Fusion

Data fusion is commonly determined as the method that composite data from various sources and accumulate this data in order to accomplish inferences, which will be more effective and possibly more precise than if they were accomplished by methods for a solitary source. This is a hop by hop congestion control protocol for traffic. Congestion detection depends on queue length. Because of transmission of wireless channel, the neighboring nodes of the congested node stops forwarding packets to the congested node and in this way remove congestion rapidly.

Congestion detection and avoidance (CODA)

Buffer occupancy is widely used as scalability in traditional congestion detection algorithms. However, the amount of buffer occupied due to the channel's shared use is not a good criterion for detecting congestion in IoT. In CODA, the receiver only checks the buffer occupancy, but measuring the channel's use in different situations is necessary to detect congestion. In general, the channel load increases much faster than the buffer occupancy during congestion. In other words, the number of packets successfully reach to the destination node. Therefore, reducing the amount of buffer occupation in the transmitter will cause incorrect information about the congestion status. For example, energy-saving occurs only when the transmission buffer is not empty.

2.2 Related work

Sankarasubramaniam, et al. [15] presented the ESRT algorithm for reliable transfer of to the destination, which focuses on the reportage measure by sources to accomplish occasion reliability with negligible resource utilization. As indicated by grouping techniques, ESRT utilizes a local congestion detection method. furthermore, the ESRT rate control method is concentrated, and resource control is performed based on essential access. ESRT utilizes a closed-loop congestion control method, in which preparing is done basically at the destination. ESRT accepts that destination nodes are sufficiently able to get to all source nodes on the network through Broadcast. The key idea in ESRT is that the destination node instructs the source nodes to change their recurrence dependent on reliable destination estimations and network congestion status. From one perspective, when the reliability of the destination node is not exactly needed, if there is no congestion on the network, the destination instructs the assets to expand the rate of their conservative revealing, if congestion in a network exists, the destination instructs resources to decrease their detailing rates. Reliability is measured at a higher destination than required, and then the destination instructs resources to reduce their reporting rates to save energy. ESRT consists of two parameters:

Reliability index, calculated by destination.
Current congestion situation.
Calculations 𝜂𝑖, in the period, the i-th is calculated by the relation

Where, the observed reliability and R_i is the reliability of the desired event at the destination. To inform the destination node about the current state of congestion, each sensor node monitors the queue's size, sets the congestion bit in the packet, and then sends it to the destination if another piece of source data causes overhead in the node queue. The rest of the parts are not destroyed. The assessment outcomes indicated that the mentioned method could can make the network more specific and reliable. Nevertheless, the mentioned technique has low load balancing and high-energy efficiency.

Swain and Nanda [16] introduce PCCP5 protocol, which is a priority-based congestion control protocol. This protocol obtains the degree of congestion by dividing the arrival time of the package and service time. In sending packets, the first priority is to node with a data. This protocol uses another method instead of AIMD to set the packet-sending rate. Nodes with a higher priority than other nodes will use more bandwidth and use more traffic to the network. Packet recovery is not possible in this protocol. Thus, PACCPS is a new priority-based rate adjustment algorithm that exists in each sensor node; it is suitable for guaranteeing and efficiency. This method has a good performance in throughput and delay. However, this method has limited functionality in energy consumption and network lifetime.

Gholipour, et al. [17] introduced a traffic-aware routing algorithm that can route packets in case of congestion. The key point of this algorithm is to use two independent fields, "Depth" and "Traffic Load”. The distance depth cost field is responsible for finding the shortest path to guide packets. The next field examines the traffic load of neighboring nodes. ; To be selected as the next node to send the package. If the traffic load exceeds the threshold, there is congestion on this route to the destination, and packages must be diverted from another route. Therefore, the traffic load field is a solution for traffic awareness, and the depth field is the routing of packets to the destination. The two fields are combined to make the decision about routing possible .This method tries to balance between choosing the shortest route and the route without congestion or less congestion. The normalized buffer size field prevents packets from entering the congestion area. In this method, the node selects one of neighbors as the next node by considering queue length. In other words, the neighbor with the shortest queue length is selected as the next node. This algorithm has a good performance in energy consumpation and packet delivery rate, but they did not pay attention to the reliability.

Razzaque, et al. [18] developed the DMQoS protocol. The DMQoS protocol's notable features are separate routing functions based on QoS parameters and modular architecture. The DMQoS protocol uses a step-by-step routing method and this method causes a data packet to have multiple choices in order to reach the destination. In addition, DMQoS is a position-based algorithm. In DMQoS, using a dynamic data classification module, packets are categorized into four types: normal data, latency-sensitive data, reliability-sensitive data, and high-sensitivity critical data. Normal data are related to periodic measurements of patients' physiological parameters, in which the issue of delay and reliability is less important. Sensitive data must reach the destination within the deadline. Reliability sensitive data must reach the destination without loss and wastage. The issue of accuracy of information is more prominent than the issue of delay. Delays and reliability are also important in critical data. The DMQoS delay control module provides a routing strategy for critical and delayed packets. In this method, the weight average function is used to estimate the delay of the node queue and the weight average function is used to calculate the transfer delay in the node.

Vaibhav Narawade proposed scheme is to avoid the congestion and to control congestion if it occurs. The congestion management in the proposed system is made possible by the adaptation of the rate adjustment optimization. The status of the congestion is active at the node means the ACSRO took over the action, rectifying the congestion in the network. Initially, the congestion avoidance on the incoming packets (traffic) from the child node is carried on by the adaptation of the packet drop. The packets with the less importance are dropped thereby avoiding the congestion. In some cases, the incoming packets cannot be dropped because of its importance criterion. There is no surety that the data packet out of the congestion avoidance is free of congestion/susceptible to congestion. When the sending rate increases considerably greater than the service rate of the parental node, then congestion occurs. If the congestion tends to happen, then the level of the congestion is found out using the total size of the queue. The congestion level corresponds to the stage at which the incoming packets start to congest. The congestion level is set within a threshold value. When the congestion level exceeds the threshold, the congestion tends to happen so that the rate optimization procedure is called. The performance of the proposed rate optimization approach is evaluated using the evaluation metrics such as throughput, delay, normalized packet loss, normalized queue size, and congestion level.

K. Kalaikumar1 proposed protocol integrates the notion of media accessibility and energy proficient hierarchical based cluster routing for increasing the network lifespan and energy efficiency. This proposed cross layer protocol uses fuzzy logic by means of considering the fuzzy descriptors such as, communication link reliability, number of neighboring nodes and precipitant (residual energy) for CH selection. Then, the network is organized into non-similar sized clusters (i.e. the clusters that are much closer to the MS holding smaller sizes and the clusters that are far away from the MS having larger sizes) for addressing the hot spots problem in WSN. The smaller sized cluster are considered because, the CHs of the smaller sized clusters are largely closer to the master station; thus, they experienced only small amount of intra-cluster congestion and forwards the relay traffic effectively using their preserved energy. Finally, the proposed FCOABC protocol employed the Oppositional Artificial Bee Colony optimization algorithm for performing inter cluster multi-hop routing from CHs to master station; therefore, an energy efficient and reliable data transfer is achieved to the master station. The authors used NS2 simulation environment developed to simulate a network. The assessment outcomes indicated that the mentioned method could decrease the energy consummation and latency; nevertheless, the mentioned technique has low throughput and low load balancing.

Pavlos Antoniou proposed approach is simple to implement at the individual node, involving minimal information exchange. In addition, it displays global self-properties and emergent behavior, achieved collectively without explicitly programming these properties into individual packets. Performance evaluations show the effectiveness of the proposed Flock-based Congestion Control (Flock-CC) mechanism in dynamically balancing the offered load by effectively exploiting available network resources and moving packets to the sink. Furthermore, Flock-CC provides graceful performance degradation in terms of packet delivery ratio, packet loss, delay and energy tax under low, high and extreme traffic loads.

This section introduces the energy consumption model and considers some settings for WSN and at the end of this section, discuss the proposed method.

3.1 Problem statement

A data center is a set of software and hardware facilities and equipment that are created to host computer and communication facilities to provide large-scale information and Internet services[19]. Due to the rapid growth of the Internet, the number of data centers is increasing day by day[20]. Some of these data centers have a bandwidth of more than 4 GB per second through different fiber-optic lines. Congestion in computer networks refers to the over-aggregation of transmitted data in the queue and the consequent loss of packets. Network congestion occurs when data packets enter the output port faster than can be transmitted and, the amount of data entering higher than the output tolerance also outputs are the largest source of congestion on the WSN system. Whenever the number of packets sent to the network is commensurate with the network capacity, all of these packets will reach their destination. Whenever the number of packets sent to the network is commensurate with the network capacity, all of these packets will reach their destination. In addition, as soon as the traffic increases excessively, the routes cannot send data, and the data packages start to get lost. This problem becomes more acute, and in high traffic, the efficiency is completely reduced, and no package will be delivered to the destination. Several factors can lead to congestion in WSN. If a sequence of packets is received on three or four input lines and the output line is all the same, queues will be formed and if there is not enough memory space to store them all, the packets will be lost. Figure 1 shows the data transfer process with and without congestion. Slow processors can also cause congestion.

3.2 Energy consumption model

Here we use the same power consumption model presented in [21]for network radio components. In this model, nJ / bit is the E_elec unit of energy used in the electronic circuit of the receiver or transmitter to receive or send one bit of the data packet. The unit E_amp (pJ / bit / mn) show the energy consumption in the WSN, and send one bit of the data packet between the transmitter node and the receiver node. Figure 2 shows the energy consumption model.

Figure (2): Model used for energy consumption in radio performance

Power consumption of the proposed model is processed dependent on the model, which was presented in LEACH method. In this model, for communicating a data packet with k bit volume and d distance d (among sending and receiving nodes), the measure of consumed energy in the transmitting node is estimated by the following formula [22]:

$${E}_{tx}\left(i\right)=k ({E}_{elec}+{E}_{amp}*{d}^{2})$$

2

Where E_elec refers to the power consumption of the electronic circuit. Additionally, Eamp signifies the necessary power for enhancing the sent and received signals so as to send a data bit. Besides, for achieving a data packet with k bit volume, the amount of power consumption for the receiving node is measured by the following equation:

$${E}_{rx}\left(i\right)=k* {E}_{elec}$$

3

İt should be noted that power consumption for the nodes located between data transmitting node and the node which sends data is equal to the sum of the energy required for sending data and the energy needed for receiving data :

$${E}_{cons}\left(i\right)= \sum _{i=1}^{n}\left[{E}_{tx}\left(i\right)+ {E}_{rx}\left(i\right)\right]$$

4

İt should be noted that the power consumed by any nodes in the network, including both data sending and data receiving node, can be measured by the above-mentioned formula.

3.3 Network model assumptions

In this paper develop the following assumptions concerning on WSN as shown: [23]:

N nodes are distributed consistently in the M*M area. Each node has a unique ID; after the establishment of the network, humans do not intervene in the network.
All sensor nodes are homogeneous, and the first energy of nodes is the align. Processing ability are equal, and energy cannot be recharge.
Every node has direct communicate with the sink. The cluster head can integrate data from nodes and send to the base station.
The energy and computing of the base station is unlimited.

3.4 Ant Colony Optimization Algorithm and Tabu search algorithm

In this section, we first introduce the ACO algorithm and the TS algorithm. Then, we present a hybrid algorithm for route detection and congestion control according to this algorithm.

3.5 The ACO algorithm is used to routing

One of the methods used to solve routing problems in WSN is the ACO algorithm. This algorithm is inspired by the behavior of ants in nature[24]. Ants do useful things such as finding the shortest path to food and sharing information with their colonial members[25]. In fact, the ant alone is a simple, unconscious being, but a set of ants can do useful things and guide humans in many applications such as navigation. The ACO algorithm has been used successfully to solve optimization problems [26]. The main purpose of the ACO algorithm is to find the path with the lowest cost in the route, and the ACO algorithm involves a large number of ants that move along the graph and find the shortest path [27]. Ants typically do not have knowledge which path is shorter, so they can only find low quality path, but global coordination among ants in a colony leads to find optimal and short paths[28]. In the real world, ants release some pheromones along the way. The decision to move ants in a path is based on the pheromone concentration in that path[29]. Ants prefer to move in a direction where the amount of pheromone is high. On the other hand, artificial ants have properties not found in natural ants. Their movement is usually consistent with their previous operations, which are stored in a special data structure. When ants move from one node to another node, they release some pheromones proportional to the path. Therefore, short and suitable paths will have a high average pheromone, and on the other hand, if the path is weak, the amount of pheromone along the path will be low. In fact, the amount of pheromone remaining in the pathway is proportional to the pathway's quality. In this way, the ants find the shortest paths in a graph. Suppose there are four ants and two paths (path R₁ and path R₂) that lead to a food, so that path R₁ is larger than path R₂ (R₁˃ R₂). There are also 6 nodes in the paths, which, as shown in the figure: Fo, N₄, N₃, N₂, N₁, N_e

At first, the four ants A1, A2, A3, A4 are in the initial choice area in Ne, and they should choose among way R1 and way R2 to arrive at food on F0.

The steps are as follows:

In the initial node ”N_e” all the ants are unaware of the positon of the food “f₀”, so they accidentally choose one between paths R₁ and R₂. Suppose ants A₁, A₂ choose path R₁ and ants A₃, A₄ choose path R₂.
A₁, A₂ pass through R₁ and A₃, A₄ through R₂, they leave a specific measure of pheromone in their way. In view of parameters, for example, the number of nodes that have been passed and the amount of path traffic, etc., the pheromone that must be placed is determined by updating the τ_Rx parameter.
Since according to the Fig. 3 (R1˃ R2), A₃ and A₄ attain F₀ before A₁, A₂.

When A₃ and A₄ follow the path R₂ to reach F₀, it passes through two nodes and sets the value of τ_R2 to 2, but A₁, A₂ have not yet reached F₀ and τ_R1 = 1. Therefore, to get back to Ne from F0, A3 and A4 again need to choose between R1 and R2 to return. In A₃ and A₄ the pheromone is τ_R, and they find that (τ_R2 > τ_R1), therefore, they choose the R2 route with a higher probability. Because the pheromone concentration is higher in the R2 pathway, and it is best solution to selected this pathway. If an ant at some point has to choose between two paths in which a pheromone has not yet been placed. It selects one path at random and with equal probability, and when the pheromone is greater than one path, that path will be more likely to be chosen.

3.6 The TS algorithm in congestion control issues

The TS method was based on local search methods and tries to fix their flaws and prevent them from falling into the optimal local answer [30]. In fact, the TS method is a localized search method that uses flexible memory structures, and it can be used as a general framework for new meta-heuristic methods [31]. This algorithm can be used for any process that provides a set of movements to change from one solution to another and uses the evaluation function to measure movements. TS algorithm has a high degree of flexibility and often creates new momentum, and uses the evaluation criteria to achieve the goal[32]. The TS method uses a concept called Tabu List to avoid falling into the trap of local optimal answers. The Tabu list contains a limited number of problem answers. Even if the best answer in the neighborhood is the current answer, this banned list will be updated at every step. Algorithm memory can be of two types: recency or frequency. Recency memory: Short-term memory in the tabu search method is an active search to find the best answers with the most desirable. It can be stated that the core of the TS algorithm is done in a short process. Frequency memory: known as long-term memory, complements recency memory by adding additional information, such as how many times a motion or Tabu answer has been searched. In this paper, the path efficiency method for topological weights is used by TS algorithm. The common link that has the highest number of passes through these routes is called the congestion link. Each source has a separate Tabu list. If the congestion rout weight is more than the threshold (70%), this link will be included in the banned list, and using that link is prohibited, thus lowering the productivity percentage of the link and preventing congestion. An incremental variable is calculated using the article before placing the congestion link in the Tabu list. The incremental variable is calculated as the second shortest path for each source passed through the congestion link. Then the obtained difference between the first shortest path and the second shortest path. In Fig. 4, A and C want to send data to destination B and D, respectively. The shortest route for A is equal to (AEFB), which costs seven. In addition, the shortest path for the origin C is equal to CEFD, and this path's cost is five. Therefore, this link is congestion EF.

3.7 Proposed method

The k-means clustering algorithm divides the data set into k subset clusters so that each subsets components have the closest distance to the center of that subset. K-means (input sample) randomly selected as the centers of the clusters. (Input samples) selects the appropriate clusters based on the minimum Euclidean distance from the centers of the designated clusters. The average of each cluster is then calculated and considered as the new center of the clusters. This operation is repeated until the centers of the clusters do not change. The criteria that should be minimized in K-means algorithm:

E _k−means = $\frac{1}{C} \sum _{k=1}^{C}\sum _{xϵ Q}\left| \right|x-{c}_{k }\left|\right|$ (5)

In the above relation, C is the number of clusters, Q is the cluster and ${c}_{k }$ is the center of the cluster.

In this simulation, the number of clusters is constant. There are several significant parameters in the selection of clusters in the formed clusters. There are three criteria for choosing a sparkler:

CH sensor should have a high-energy level; Because the CH has higher overhead than other nodes, the node must be selected as the CH with enough energy; otherwise, the node will be disconnected from the base station due to the death of the node.
The CH sensor closest to the base station; the less energy it needs to send data packets.
The CH sensor closest to the center of the area; In fact, it is a sensor that the average distance between other nodes is minimal.
In fact, the centrality of the CH reduces the energy consumption for communication within the other nodes.

Data transfer stage

After the CH is selected, it is time to transfer the data. In this case, we have a series of inter-cluster and intra-cluster communications. In sending data, the normal node sends the data to the CH, and then the data is placed in the best path and sent to the sink station. This section uses the pheromone and the possible path selection function. To perform this step, we defined several messages, which are as follows.

Hello message:

Hello packets or routing packets are sent periodically as well as at the first start of the algorithm to all neighbors, so that the neighbors are known. When neighbors receive these packets, they will confirm their response with the packets. During the message sequence, these bandwidths as well as the path pheromone are updated.

Route request message:

When we have a path to send data, the path request packet or the ants ahead will be created to identify the paths. In all identified paths, nodes are seen along the way, as well as the number of steps to reach the destination, latency, energy stored in an array by routing ants.

Route message

As soon as the route request message arrives at the destination, the route request becomes the route response to ant packet. The path requested by the ant packets will be reversed, the stack node will be emptied and the path parameters will be updated.

Wrong path message

When a node concludes that it is not reaching the destination. The wrong path message is sent and, defining such a massage will reduce the loss rate of packets.

Possibility to choose the route

The probability of selecting the path from the source node i to the destination d is calculated according to the following formula, given that node j is adjacent to the node i:

P_ijd = $\frac{\left({{\tau }}_{ijd}{)}^{{\alpha }}\right({\text{D}}_{ijd}{)}^{\text{ß}}\left({\text{ŋ}}_{ijd}{)}^{\text{ß}}\right({BP}_{ijd}{)}^{\mathcal{s}\mathcal{ }}({E}_{ijd}{)}^{\text{k} }}{\sum i ϵ{N}_{i} \left({{\tau }}_{ijd}{)}^{{\alpha }}\right({\text{D}}_{ijd}{)}^{\text{ß}} \left({\text{ŋ}}_{ijd}{)}^{\text{Ŷ}} \right({BP}_{ijd}{)}^{\mathcal{s}\mathcal{ }}({E}_{ijd}{)}^{\text{k} }}$ (6)

𝝰, ß, k, Ŷ, τ are weight regulators, which we consider here are related to energy, bandwidth, latency, number of steps and pheromones. We considered the number of steps as a heuristic function; also, D_ijd shows bandwidth and E_IJD as the energy parameter.

Pheromone

Initial pheromones are zero because no neighbors are known, and when they are known together, the initial pheromone changes slightly. When node j is known as neighbor i, this value changes to 0.1. When the path message is sent from node i to j the τ_ij changed to the 0.1, This means that there is a path to destination (d) using the link (I, j), so the pheromone of this path must be updated. We have updated this pheromone based on the following relation.

τ_{ij =} τ_{ij + Δ} τ_ij (7)

In the above relation we considered the amount of pheromone to be constant, although I can consider this relation as a function based on delay or bandwidth. In the above relation, we considered the amount of pheromone changes to be 0.5.

0.5 = _Δ τ_ij (8)

The idea of using a pheromone is based on path if a data transfer is appropriate, it will be used over time. On the other hand, we need a parameter that if a communication path is lost or a node is deleted, or data is not transmitted to the desired neighbor, the path is no longer amplified not to be selected in subsequent communications. That is why the pheromone evaporation factor is used here. This factor is called ρ. We include this value in τ_ij as follows.

$${\tau }\text{i}\text{j}= \left\{\begin{array}{c} \left(1-p\right)\tau ij if \left(1-p\right) \tau ij>0.1\\ 0 if \left(1-p\right) \tau ij<1\\ 0.1 otherwise\end{array}\right.$$

9

According to the above formula, when node i loses the connection with node j, the pheromone of that link will be zero.

Path discovery phase

Ants move from S to D. This move is due to the neighbors who are known by the message of greetings. When the ants move to the destination, they store path information such as latency, number of steps, available capacity of each link. When the route request packets reaches to destination, the route response packets is created, and these packets follow exactly the same path as the requested packets. Therefore, each node can get the best chance of reaching the destination.

Path maintenance phase

When data is transmitted, the path with the most amplification value and the greater probability of selection is considered the best path. In addition, when a route is always chosen to send, we will have the problem of delay. Nodes may lose energy due to the selection of fixed nodes to send data. In this method, because this parameter is considered in the probability function, the selection of this path is automatically reduced in the next steps and other paths are replaced.

After discovering the best path ,the TS algorithm starts.
Calculate the path weight by topology

In this step, the productivity is calculated by the weighting method for the routes. If one node is not connected to another node, it gets a value of zero. Otherwise, it contains a number that shows the percentage of productivity of that link. If N nodes exist in the network, the route from source to destination is obtained so that congestion is controlled.

Find the most data transfer route.

A congestion matrix is formed whose dimensions are equal to "one". In the congestion matrix, the number of passes per link is placed. The array, which contains the largest number in this matrix, this link has the highest number of transfer in path.

Calculate the threshold of the congestion path

In this step, the obtained link is checked. Is the link weight more than 70%. If the weight of this link is more than 70%, it will be included in the congestion list. Otherwise the link is ideal. In the formula 5 X_ij means selecting the link i to j. if this link is selected the X_ij =1 and if this link is not selected X_ij =0. In the event that X_ij is not selected the node I to j is not has link. The congestion parameter is called C_ij. These parameters show the input data to the link capacity. The purpose of this problem is to find the best path called Z.

Minimize Z = $\sum _{i=1}^{n}\sum _{j=1}^{n}{C}_{ij }{X}_{ij}$ (10)

Subject to :

${X}_{ij}$ = 0 or 1 ( I = 1,2,…, n ; j = 1,2,…,n) (11)

Check route efficiency

In this method, the route is checked according to the efficiency of the route and the data reaching the CH. also, all nodes send their own data to the CH.

Add to Tabu list

Each source node has a Tabu list. For routes that have passed through the congestion link, one of them is accidentally left out. That is, the banned list of the desired source node does not change. But for other routes, the congestion link is on the banned list. This means that it can not send data from this link.

Routing again with the help of ACO algorithm

After discovering the congestion route, the same nodes are routed again.

Using an identical dataset, we investigated and compared the performances of the proposed method by ACSRO algorithm [33], FCOABC algorithm [34], and Flock-CC algorithm [35]. Simulation results of the proposed algorithm with the above-mentioned algorithms were scrutinized.

4.1 Simulation environment

The features, settings and the used information were simulated in R2016b NS3 software. The applied hardware in this study had the following specifications: Intel core I 7 processor, 2 GB RAM memory and Windows 10. It should be noted that NS3 software is generally used for evaluating wireless networks and IoT. Most researchers prefer to use NS3 for simulating and modeling algorithms. NS3 software has numerous mathematical merits over the other software which makes it possible to share source codes among users. Given these merits and advantages, we decided to use NS3 software for simulating the proposed method.

4.2 Simulation parameters

We used the Simulation parameters introduced in with respect to simulation parameters production. As shown in Table 1, for facilitating and simplifying the simulation procedure, we limited the following parameters: network size, number of sinks, number of nodes, data packet size, initial power of each node, buffer size of each node and the number of iterations.

Table 1

Simulation parameters [21]
Network parameter	Values
Number of base stations	1
Number of nodes	200
Data size of packet	2000 bits
packet size	25 bits
Intial energy of node	1 j
Initial buffer size of node	65 Kbyte
Number of rounds	1500
𝐸_{𝑒𝑙𝑒𝑐}	50 nj/bit
𝐸_𝑎𝑚𝑝	0.0013 pj/bit/ m²

Table 2

Parameters of scenario [23]
Parameter	Scenario 1	Scenario 2	Scenario 3
Area	100 * 100 m²	200 * 200 m²	300 * 300 m²
Sink area	;50 50	0;0	300;450
numeral of node	(100,200,300)	(100,200,300)	500

Table 3

The parameter of Tabu algorithm
Parameters	Value
Simple edge length	0.1*min
Ratio of accepting diversification point	1
Maximum of number iteration	200
Termination accuracy	10^− 6
Number of Tabus	5 * number of variables
TR radius	0.01 * SEL
Length of tabu list	100
Length of candidate list	50

Table 4

The parameter of ACO algorithm
Parameters	Value
Number of ants	100
Pheromone exponential weight	1
Heuristic exponential weight	1
Evaporation rate	0.1

4.3 The obtained results

In this part of the research, we review the results of practical experiments. The diagrams obtained for all three scenarios examined in different networks FCOABC algorithm, Flock-CC algorithm, and ACSRO algorithm, are included. In fact, in this section, we have obtained the result of comparing the performance of different networks in three scenarios in terms of the number of a live nodes, network life time, network stabilization time and Throughput. It should also be noted that in comparison all parameters are the same. In the first scenario, the four algorithms are compared together; the network lifetime. Figure 6 present that the proposed algorithm has superior efficiency over different algorithms. In the primary situation with 100 sensors, 200 sensors and 300 sensors, the proposed algorithm has a more extended life. The size of the first situation is little and the sink is situated in the middle of the zone, and the distance from the member nodes to the CH and CH to sink is generally close. For this situation, the main factor influencing the presentation of the protocol is the routing effect. In view of the investigation, the fundamental of purpose for the proposed strategy to have a superior network lifetime is to choose the correct path and upgrade congestion control.

In the first scenario from Fig. 7, shows the network stabilization time. As demonstrated in the figure, in the first situation, the node congestion is various, and the network stability in the proposed algorithm are superior than the other three algorithm. Due to the proposed method can, show the balance between energy consumption, and the proposed algorithm is more appropriate for these situations with high-reliability necessities in compared to the other three algorithm.

Figure 8 shows the throughput measurements for the first situation. It is evident from this figure during the operation of the network, the sink gets more data, and in fact, the proposed technique will have a more network lifetime and more throughput. In the intra-cluster connection phase, the ACO algorithm is presented, making full use of the time gap and improving network throughput.

Figures 9, 10 and 11 show the live nodes in the first scenario with 100, 200, 300 sensors, respectively. In the first scenario with different algorithm densities, the number of live nodes of the proposed technique is better than the other algorithms. This technique demonstrates that the proposed algorithm can effectively in energy consumption and improve energy efficiency.

In Scenario 2, four algorithms are simulated in NS3 simulator. In examination with Scenario 1, the base station in Scenario 2 is situated at the side of the area, and the scale is bigger. With respect to number of sensor nodes, in this part using 100, 200, and 300 sensors. Figure 12 demonstrate the network life- time criteria in the second scenario.. As expected from the proposed method the network life compared to the other three protocols has a better performance. The reason for this is that the proposed protocol can logically route and make the CH locations more scattered.

Figure 13 depicts the network stabilize time for situation 2. The proposed method has a better network stabilize time in compered with other algorithm. The proposed strategy effectively chooses the routing and congestion control way finds the ideal path from each CH to the base station.

Figure 14 shows the second Scenario in the network throughput. According to figure, the proposed technique has ideal network throughput. The two primary cause that the proposed strategy performs better than the compared protocols. The first explanation is that the proposed method has a more extended network life, so the base station gets more data packets. In addition, the second explanation is that the way recognition mechanism is presented in the intra-cluster communication stage.

Figures 15, 16 and 17 show the number of a live node in the second scenario with 100, 200 and 300 sensors. As described in figure, the proposed algorithm carry out well in the residual energy evaluate. Routing and congestion control in the proposed method considered energy consumption. Therefore, in terms of energy consumption and energy balance, it finds the optimal solution. It also works faster in finding high-traffic routes that do not waste energy.

In Scenario 3, the network area getting expands, the quantity of sensor nodes increases to 500, and the sink is situated outside the zone. Figure 18 shows the longevity criteria for scenario 3. As depicted in figure, the proposed protocol performs better on a large-scale than the other protocols in the network lifetime criteria.

Figure 19 shows the lifetime criteria for large-scale. The proposed algorithm gives efficient inter-cluster routing and prevents over-utilization of CH energy. So it greatly extends the network stabilization time. This proposed method enables gathering more complete data, and it is reasonable for large-scale with high-reliability necessities.

Figure 20 shows the network throughput metrics for Scenario 3. From this figure, it can be seen that the throughput of the proposed method is much higher than the compared protocols.

Figure 21 shows the number of live nodes in each round of the three protocols in Scenario 3. From the figure it can be seen that the total residual energy of the proposed protocols is always more than the compared protocols. The proposed routing algorithm can solve the problem of optimal routing and prevention of congestion.

Due to the inherent limitations and the number of nodes distributed in WSN, the number of production packets is too much and congestion and traffic will occur at the communication level. Too many packages will be lost due to congestion, and network resources, including energy, will be wasted. In this method, first, using the ACO algorithm, an attempt to find the best and most ideal way to deliver data from the nodes to the CH. Then using the pheromone evaluation, the obtained path can be saved, and after discovering and saving the path, the congestion control method for WSNs will be started. Where a parameter called tabu search is used to control congestion. Discover the routes that have the most transactions among other routes, and by assessing their threshold, they can be placed on the banned list, and in case of congestion, another route is used to send packages. The simulation results indicate that the proposed method has energy efficient, and the packet loss rate and network lifetime are less than the other method. In addition, to be explored in future research can be the use of machine learning and deep learning techniques for spontaneous learning, so it can learn and apply congestion control in other ways.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Acknowledgments

This study has been conducted under the project ‘Mobility and Training foR beyond 5G ecosystems (MOTOR5G)’. The project has received funding from the European Union’s Horizon 2020 programme under the Marie Skłodowska Curie Actions (MSCA) Innovative Training Network (ITN) having grant agreement No. 861219.

Conflicts of Interest

The authors confirm that there is no conflict of interest in this research paper.

N. Dutt, A. Jantsch, and S. Sarma, "Toward smart embedded systems: A self-aware system-on-chip (soc) perspective," ACM Transactions on Embedded Computing Systems (TECS), vol. 15, no. 2, pp. 1–27, 2016.
S. A. A. Elmustafa and E. Y. Mujtaba, "Internet of things in Smart Environment: Concept, Applications, Challenges, and Future Directions," World Scientific News, vol. 134, no. 1, pp. 1–51, 2019.
P. Jadhav and V. Deshmukh, "Forest fire monitoring system based on ZIG-BEE wireless sensor network," International Journal of Emerging Technology and Advanced Engineering, vol. 2, no. 12, pp. 187–191, 2012.
I. Mosavvar and A. Ghaffari, "Data aggregation in wireless sensor networks using firefly algorithm," Wireless Personal Communications, vol. 104, no. 1, pp. 307–324, 2019.
H. Alemdar and C. Ersoy, "Wireless sensor networks for healthcare: A survey," Computer networks, vol. 54, no. 15, pp. 2688–2710, 2010.
J. Rischke, P. Sossalla, H. Salah, F. H. Fitzek, and M. Reisslein, "QR-SDN: Towards Reinforcement Learning States, Actions, and Rewards for Direct Flow Routing in Software-Defined Networks," IEEE Access, vol. 8, pp. 174773–174791, 2020.
B. Hull, K. Jamieson, and H. Balakrishnan, "Mitigating congestion in wireless sensor networks," in Proceedings of the 2nd international conference on Embedded networked sensor systems, 2004, pp. 134–147.
S. L. Yadav and R. Ujjwal, "Mitigating congestion in wireless sensor networks through clustering and queue assistance: a survey," Journal of Intelligent Manufacturing, pp. 1–16, 2020.
A. A. Rezaee and F. Pasandideh, "A fuzzy congestion control protocol based on active queue management in wireless sensor networks with medical applications," Wireless Personal Communications, vol. 98, no. 1, pp. 815–842, 2018.
M. A. Jan, S. R. U. Jan, M. Alam, A. Akhunzada, and I. U. Rahman, "A comprehensive analysis of congestion control protocols in wireless sensor networks," Mobile networks and applications, vol. 23, no. 3, pp. 456–468, 2018.
H. Haile, K.-J. Grinnemo, S. Ferlin, P. Hurtig, and A. Brunstrom, "End-to-end congestion control approaches for high throughput and low delay in 4G/5G cellular networks," Computer Networks, p. 107692, 2020.
A. Bohloulzadeh and M. Rajaei, "A survey on congestion control protocols in wireless sensor networks," International Journal of Wireless Information Networks, pp. 1–20, 2020.
I. U. Haq et al., "E2-MACH: Energy efficient multi-attribute based clustering scheme for energy harvesting wireless sensor networks," International Journal of Distributed Sensor Networks, vol. 16, no. 10, p. 1550147720968047, 2020.
O. Ahmed, F. Ren, A. Hawbani, and Y. Al-Sharabi, "Energy Optimized Congestion Control-Based Temperature Aware Routing Algorithm for Software Defined Wireless Body Area Networks," IEEE Access, vol. 8, pp. 41085–41099, 2020.
Y. Sankarasubramaniam, Ö. B. Akan, and I. F. Akyildiz, "ESRT: Event-to-sink reliable transport in wireless sensor networks," in Proceedings of the 4th ACM international symposium on Mobile ad hoc networking & computing, 2003, pp. 177–188.
S. K. Swain and P. K. Nanda, "Priority based fairness rate control in wireless sensor networks," in 2017 International Conference on Wireless Communications, Signal Processing and Networking (WiSPNET), 2017, pp. 2206–2210: IEEE.
M. Gholipour, A. T. Haghighat, and M. R. Meybodi, "Hop-by-hop traffic-aware routing to congestion control in wireless sensor networks," EURASIP Journal on Wireless Communications and Networking, vol. 2015, no. 1, p. 15, 2015.
M. Razzaque, C. S. Hong, and S. Lee, "Data-centric multiobjective Q-aware routing protocol for body sensor networks," Sensors, vol. 11, no. 1, pp. 917–937, 2011.
E. S. Grange et al., "Responding to COVID-19: the UW medicine information technology services experience," Applied clinical informatics, vol. 11, no. 2, p. 265, 2020.
P. Huang et al., "A review of data centers as prosumers in district energy systems: Renewable energy integration and waste heat reuse for district heating," Applied Energy, vol. 258, p. 114109, 2020.
H. Mosavvar and A. Ghaffari, "Detecting Faulty Nodes in Wireless Sensor Networks Using Harmony Search Algorithm," Wireless Personal Communications, vol. 103, no. 4, pp. 2927–2945, 2018.
H. Parsavand and A. Ghaffari, "Controlling congestion in wireless sensor networks through imperialist competitive algorithm," Wireless Personal Communications, vol. 101, no. 2, pp. 1123–1142, 2018.
Z. Wang, H. Ding, B. Li, L. Bao, and Z. Yang, "An energy efficient routing protocol based on improved artificial bee colony algorithm for wireless sensor networks," IEEE Access, vol. 8, pp. 133577–133596, 2020.
H. Zang, S. Zhang, and K. Hapeshi, "A review of nature-inspired algorithms," Journal of Bionic Engineering, vol. 7, no. 4, pp. S232-S237, 2010.
M. Dorigo, M. Birattari, and T. Stutzle, "Ant colony optimization," IEEE computational intelligence magazine, vol. 1, no. 4, pp. 28–39, 2006.
Z. C. S. S. Hlaing and M. A. Khine, "Solving traveling salesman problem by using improved ant colony optimization algorithm," International Journal of Information and Education Technology, vol. 1, no. 5, p. 404, 2011.
M. P. Garcia, O. Montiel, O. Castillo, R. Sepúlveda, and P. Melin, "Path planning for autonomous mobile robot navigation with ant colony optimization and fuzzy cost function evaluation," Applied Soft Computing, vol. 9, no. 3, pp. 1102–1110, 2009.
Y.-N. Ma, Y.-J. Gong, C.-F. Xiao, Y. Gao, and J. Zhang, "Path planning for autonomous underwater vehicles: An ant colony algorithm incorporating alarm pheromone," IEEE Transactions on Vehicular Technology, vol. 68, no. 1, pp. 141–154, 2018.
D. Arulanantham and C. Palanisamy, "DACASN: Dynamic ant routing-based channel accessing cognitive sensor network in Internet of Things," International Journal of Communication Systems, vol. 34, no. 1, p. e4672, 2021.
M. Alinaghian, E. B. Tirkolaee, Z. K. Dezaki, S. R. Hejazi, and W. Ding, "An augmented Tabu search algorithm for the green inventory-routing problem with time windows," Swarm and Evolutionary Computation, vol. 60, p. 100802, 2021.
J.-F. Cordeau, G. Laporte, and A. Mercier, "A unified tabu search heuristic for vehicle routing problems with time windows," Journal of the Operational research society, vol. 52, no. 8, pp. 928–936, 2001.
F. Glover and M. Laguna, "Tabu search," in Handbook of combinatorial optimization: Springer, 1998, pp. 2093–2229.
V. Narawade and U. D. Kolekar, "ACSRO: Adaptive cuckoo search based rate adjustment for optimized congestion avoidance and control in wireless sensor networks," Alexandria engineering journal, vol. 57, no. 1, pp. 131–145, 2018.
K. Kalaikumar and E. Baburaj, "Fuzzy enabled congestion control by cross layer protocol utilizing OABC in WSN: combining MAC, routing, non-similar clustering and efficient data delivery," Wireless Networks, vol. 26, no. 2, pp. 1085–1103, 2020.
P. Antoniou, A. Pitsillides, T. Blackwell, A. Engelbrecht, and L. Michael, "Congestion control in wireless sensor networks based on bird flocking behavior," Computer Networks, vol. 57, no. 5, pp. 1167–1191, 2013.

No competing interests reported.

Routing and Congestion Control in Wireless Sensor Networks using the Ant Colony Optimization algorithm and the Tabu search algorithm

Status:

Version 1

Abstract

Figures

1. Introduction

2. Background And Related Work

2.1 Congestion Control Strategy

2.2 Related work

3. System Model

3.1 Problem statement

3.2 Energy consumption model

3.3 Network model assumptions

3.4 Ant Colony Optimization Algorithm and Tabu search algorithm

3.5 The ACO algorithm is used to routing

3.6 The TS algorithm in congestion control issues

3.7 Proposed method

4. Experimental Results

4.1 Simulation environment

4.2 Simulation parameters

4.3 The obtained results

5. Conclusions And Future Work

Declarations

References

Additional Declarations

Status:

Version 1