An Improved Role Based Trust Management System Using Interactive Artificial Bee Colony (i-abc) Algorithm for Wireless Sensor Networks

The aim of this research study is to propose an improved role based trust management system for wireless sensor networks. The objective is attained using an interactive artificial bee colony algorithm which is used for path optimization. Also the purpose is to guarantee to provide robust and trustworthiness among the clusters formed in the wireless sensor network. The reputation among the sensor nodes is assured using the trust model. One of the main concerns of WSN, that have attracted research scholars, is the property to guarantee a less amount of security in a limited environment. Trust among the communicative nodes is one of the major issues that have to be given importance in wireless sensor network. A number of research works have concentrated on only Trust Management (TM) techniques without considering their roles. Existing trust management schemes do not provide significant reliability in all environments. In order to overcome these issues, Role Trust (RT) framework is presented, to select role for each nodes representing policies in a distributed authorized environment. RT integrates the features of role-based access control and TM schemes. This feature is particularly applicable for attribute based access control. Role trust management schemes generate rules, identify and assure roles based on the rules generation process and then clusters are identified for each trust model. Along with the trust model, this study focuses on efficient data path with reduced data loss. Hence, this study presents a novel Swarm Intelligence based Role Trust and Reputation Model (SIRTRM) to provide trust and reputation in WSNs. Simulations are carried out in NS2 environment and the results portrays the accuracy, robustness and lightness of the proposed model.


INTRODUCTION
WSNs consist of several spatially distributed components that observe and transmit the aggregated data to a position in middle with the help of wireless channels.Since restricted resource facilities of WSN, it is not an easy task for integrating fundamental security model such as authentication, reliability and key distribution.Since WSN have several interesting science and engineering applications, a number of research works have been carried out to develop efficient security schemes.But, most of the WSNs consist of nodes with highly dynamic topology which may results in certain scalability issues.
As a result, WSN are susceptible to various types of malicious attacks, such as denial of service, routing protocol attacks as well as replay attacks.Conventional crypto schemes are insignificant in handling such malicious attacks.Thus, trust and security management schemes have become an attractive research area in WSN (Aivaloglou et al., 2006).However, conventional trust management approaches may not be appropriate for networks with small sensor nodes due to restricted bandwidth and rigorous node constraints in terms of power and memory (Ganeriwal and Srivastava, 2004;Yao et al., 2006).Moreover, it is observed from the literature that certain techniques do not consider the dynamic factors of trust (Shaikh et al., 2009) which results in insignificant performance of dynamic topology.
Another major issue in WSN is that nodes could misbehave when certain data are extracted from them.Conversely, the sensor nodes are usually fixed in environment where there is a possibility of an intruder are able to capture them along with consequently use these nodes to attack the whole network (Ren et al., 2003).Therefore, it is necessary to recognize the compromised nodes and eliminate them from the network so as to prevent the whole network from the intruders.
This research study aims to develop a more suitable role based efficient trust model for WSN in order to enable mobile sensor nodes to exchange trust opinions with minimal overheads.To represent policies for each node such as either cluster head or cluster member and credentials in distributed authorization.Development of the RT framework is the main effort to address security issues that arises when independent organizations enter into coalitions and for system whose membership and existence change rapidly.
In general, client server architecture depicts some nodes of the network request some services and others provide those services.Sometimes a node could provide a fake service for triggered request.In order to avoid such situation, this research work uses the best path selection based on swarm intelligence approach.Swarm Intelligence based Role Trust and Reputation Model (SIRTRM) is proposed in this research for the selection of the most trustworthy node based on the reputable path selection offering a certain service.The swarm intelligence used in this approach is Artificial Bee Colony (ABC) Algorithm.

LITERATURE REVIEW
Security considerations in WSN have increased in the last two decades mainly due to its applications in various military related applications.Trust based management system have increased to overcome the problem of security level in WSN (Ferraiolo et al., 2001).Most of the earlier trust management systems do not consider trust level for individual nodes in the network along with roles.For example, a multi path routing protocol based on dynamic clustering methods and ACO, is described in MRP (Yang et al., 2010), which enhances better data aggregation with trust, thereby reducing the energy consumption.
Fuzzy and extended fuzzy based trust management methods for WSN have also been proposed for effective management.Zarei et al. (2009) proposed a fuzzy logic based framework to control network traffic in WSNs.TFCC protocol is proposed in Chakaborty et al. (2013), which controls congestion traffic of network, from source nodes to sink nodes with improved network throughput.Trust based management system for WSN uses a swarm intelligence based methods (Dhurandher et al., 2009) to improve reliability and QOS in WSNs.QoS results are estimated based on Quality Distance Vector (QDV) protocol.The result of this shows that the proposed swarm intelligence based protocol have to achieve higher reliability for communication.The above mentioned trust management methods does not assign roles for each nodes in WSN.In such methods, the nodes are randomly chosen as cluster heads.Based on chosen cluster head, remaining nodes are grouped.Thus, there are still spaces for better security schemes.Therefore, the present research work focuses on role assignment to each node in the network using role based languages.RBAC Model (Ferraiolo et al., 2001) provides flexibility category of access control procedure.An access control policy group similar user roles and allows users to access the resources of the system.The entire description of role trust language specification for various applications is described in Li and Wang (2007) and Li et al. (2003).In this study, based on role trust framework, roles are assigned to nodes in WSN.

METHODOLOGY
Proposed Role Trust Management System (RTMS) and optimal path selection using Artificial Bee Colony (ABC) algorithm: In earlier work, (Marmol and Perez, 2011) developed an efficient trust model framework for WSN based on bio-inspired optimization methods.It follows the procedure of ant colony system based on user requested service in WSN, Where each ant is considered as nodes that receives the right requested service and gets adapted, to find the trust worthiness of a cluster head.The results are carried out to measure accuracy and strength of this trust model.The present research work, trust management system is extension of this model and it is enhanced in several points of view.The existing trust based management system does not consider the roles based on the efficiency of the rules.The proposed system creates a role trust based cluster model in WSN to select roles such as Cluster Head (CH) and Cluster Member (CM) along with the residual energy and trust value for each node.Once the cluster head is elected, then the next process is the formation of clusters based on the Role based Trust with semantic Rules (below section).Based on the semantic rules, the most important trusted node is elected as a cluster head.Which forwards the data to the selected cluster head and also an efficient path selection is done to destination node using ABC without loss of packet.Then, ultimately, the current cluster head energy level is checked with specified threshold value, if less than threshold value then reelection are triggered.The proposed system is shown in Fig. 1.
The basic concepts of roles are analyzed by the role based trust management system model and there by the trust value of each node are computed.In this study, a node to node level trust is considered and it is estimated based on the earlier interaction among nodes in communication.Similarly, trust values are calculated for entire role nodes in role trust management.The procedure is explained below.
Cluster formation: A clustering architecture offers fault tolerance along with network scalability that will give the outcome of exploiting the network resources.Clustering can be applied effectively for resource management, routing and location management and also drastically lessen both communication overhead and computational time complexity.

Cluster head selection:
In first phase of the proposed system form a cluster by selecting the cluster head The weight value of the communication nodes w G is defined as: Similarly w I is also defined.This mapping of results improves trust values results based on assigning high, low and medium communication time.The time t G is defined as: Peer recommendations evaluation: Assume that there is n number of nodes in WSN and every node in WSN has a unique id.When any node gets peer suggestion then it calculates the trust value of each node in the network by using following formula: where, ˜˞ , = The peer recommended trust value of node y calculated by node x ˠ˰ , = The trust value of node 'i' calculated by node x ˠ˰ , = The trust value of node 'y' calculated by node i Let's suppose node x wants to calculate the trust value ˠ˰ , of node y, then it can be evaluated by the following equation: After the trust values are calculated for each node, then the process of creating the clusters by assigning the roles for each node.Clustering is done for each node after the election of cluster head.Then the roles are assigned which alternatively based on the size.

Roles created based on the trust value calculation:
In general, role based trust management system follows the procedure of creating roles (Li and Mitchell, 2003), it can be implemented to cloud computing methods to access control of VMs (Lasota and Kozakiewicz, 2012) and trust in WSN.Some conditions are defined to each role in the Wireless sensor networks, for discriminating roles activations.It uses various entities to differentiate roles of each node in WSN.Roles define a set of entity which is a member of either the cluster head or the cluster member.An entity is an association of a cluster head condition and only if all the nodes in the remaining trust values are comparable to the cluster head.In earlier work, ref (Gorla et al., 2006) is based on time dependent principles for role trust.A central organizing notion in RT is the notion of roles cluster head and cluster member.Each RT management principal has its own given name space for roles such as (Cluster Head, Cluster Membership), similar to localized name (highest trust values) spaces in trust management.For example, if H is a principal that is highest trust values and R is a role term of either Cluster Head (CH) or Cluster Membership (CM), then H . R is the role R defined by principal H with highest trust for each nodes and can be read as H R role.Only H can issue policy statements defining the role H . ˞.
These statements determine members of H . ˞.For example, H can define H . ˞ to include another principal H role, effectively delegating some control over the role H R to H .
A cluster group is clearly a set of principles that has the highest trust values with less communication time for node to node communication.An identity is a set of principles corresponding to one physical user (Cluster Head (CH) to Cluster Member (CM)) some systems require the set to contain just one principle.To remove suspicious node in the network that are trustworthy (Crosby et al., 2006), Cluster head selection is important in WSN.A Role in RBTM can be viewed as a set of highest trust values that are members of this role.The role hierarchy relationship that H . ˞ is more powerful than H . ˞ can be viewed as defined that all members of H . ˞ are also members of H . ˞. Permission corresponds to a set of highest trust values given by the authorities.The highest trust amounts to each node in wireless sensor network set as a principal as each node of the set corresponding to the permission for each node in wireless sensor networks.Granting permission to a role (Cluster Head (CH) and Cluster Membership (CM) amounts to asserting that the cluster corresponding to the permission which a subset the set is corresponding to the roles Cluster Head (CH) and Cluster Member (CM).To understand our concept, first define the role and principals for each roles in wireless sensor networks is based on the trust value that defines a rules to assign roles for each node, then formation of cluster.After that perform optimal path discovery for each clusters to eliminate malicious node.
Creating the roles for cluster head selection and cluster formation: Sample calculation for role trust system: The role trust system consists of basic elements such as entities, role names, roles and policies.An entity represents number of nodes in the WSN that can characterize roles as Cluster Head (CH) and Cluster Member (CM).Role names represent permissions that can be concerned by one entity to other entities, or groups of entities.Highest trust value result considered as cluster head and the remaining nodes is considered as Cluster Member (CM).Roles represent sets of nodes that have particular permissions granted according to the access control policy for Cluster Head (CH) and Cluster Members (CM).Credentials define roles by selecting a new cluster member of the Cluster Head (CH) role by delegating authority to the members of other roles.Based on the Credentials, the Cluster Head (CH) and Cluster Members (CM) are selected and cluster creation is completed: If it satisfies H . ˞1 then cluster head members are added to cluster H . ˞.The time taken for communication are converted into descending order to formation of cluster, because if the present cluster head reaches dead state based on energy it automatically selects the second lowest time communication as cluster head.Three nodes are formed a cluster and added to H . Then remaining two nodes again perform the same policies (1) and ( 2).Next it performs the ABC algorithm is performing to select path in each cluster.

OPTIMAL PATH SELECTION USING ARTIFICIAL BEE COLONY (ABC) ALGORITHM
Earlier work (Karaboga and Akay, 2011) pertains to the development of optimized framework related to ABC and solves many real time optimization problems.The Algorithm replicates the intellectual searching behavior of honey bee swarm.In ABC algorithm, the gathering of artificial bees is considered as nodes where each node in cluster is divided into three types of bees: employee bees, onlookers and scouts.Each node in the cluster performs wait based on the dancing area for the best optimal path selection.Clustering is measured as optimal path and this bee is named as onlooker.The best optimal path appointment is considered as employed bee.The remaining bee in the WSN is named as scout bee that carries out random exploration for discovering new optimal path sources.This results in the assigning of optimal value in each cluster in a cluster role trust management system.The position of each node in a optimal path represents best optimal solution and the nectar amount is analyzed to calculate the quality (fitness) of the equivalent optimal path, it is calculated by using (3): Each node sends a data to other nodes in the cluster, from which fitness values is estimated.In this study fitness function is evaluated based on weight function of a small number of parameters, like degree of the node in cluster, received power level for each node, trust factor and battery level.Trust values estimates the extension of earlier work (Bednarczyk and Gajewski, 2013).This parameter is calculated using the formula: where, ˣ # , ˣ $ , ˣ % , ˣ & are weight values of the parameters like degree of the node in cluster, received power level for each node, Trust factor and battery level.The sum of the weight values should be equal to one.Change of weight parameter values can be continuously applied to several areas: Degree of the node (DG), as a number of edges of a given node, the Received Power level (PR), Trust Factor (TF), Battery Level (BL), as a remaining work time.
In the ABC algorithm, the nodes in the clusters are separated into two bee's employee and onlooker bee's.The first half of the nodes in the cluster are the employee bees and the second half of the nodes in cluster continues to be the onlooker's bee.The total number of employee bees in the cluster is equal to the best optimal path nodes in the cluster population.Initially the best optimal path is assigned to zero and each optimal path selection results in ˴ where ˩ = 1, 2, . . ., ˟˚ is a D-dimensional vector pertaining to the number of nodes in WSN.Corresponds to the cluster size for each group is initialized to the cluster population in WSN.The best optimal path of the location is selected and is focused to repeat based on the Maximum Number of iterations (MCN).An employed bee produces the best optimal path selection result and updates its memory based on current optimal path selection which results in the RTMS cluster.Corresponding updates in position of each employee bees is done (3).If they found optimal path selection results in cluster having better than the old selection optimal path than old cluster is replaced by the new one as best optimal path else the current optimal path as new one.All employed bees (nodes) in the cluster complete the optimal path selection.As they share the nearest nodes information with best optimal path selection.Onlooker bee evaluates optimal path selection resulting in the clustering of the employee bee.It then selects an optimal path with highest probability values related to new optimal path, Provided that the nectar node results in the path selection that has higher value than the earlier.The new employee bee nodes results in the best path selection in RTMS.Every node has a collection of probability values associated with that optimal path.Food sources for all neighbors' nodes and the cluster size will find out the probability of bees selecting a definite route or another definite route.Consequently, the food source is treated as the unit of trust.Yet, in order to differentiate between food source and trust value, ˠ is defined as the trust value of node i, where 1 ≤ ˩ ≤ ˫.Trust value will be calculated based on food source from fitness function (3).An artificial onlooker bee chooses an optimal path selection depending on the fitness function (3).That corresponding positions are updated to each employee bee which results in the best optimal path selection with highest probability values J to each nodes, it is calculated by the following expression: where, SN is the number of cluster size equal to the number of employed bees present in current cluster and ˦˩ˮ is defined below: In order to find best optimal path selection in the cluster with old memory path selection result, the ABC uses the following expression (5): where, k ∈ {1, 2, . . ., SN} and j ∈ {1, 2, . . ., D} are randomly chosen indexes for optimal path selection in cluster.Although k is defined as random number it is varied from ˩. is a random number between (-1, 1).This will result in reducing the production of neighboring path optimal sources around ˴ and represents the comparison of two optimal path positions visible to a bee.In Eq. ( 6), as the difference between the position of different nodes in cluster ˴ , and ˴ , decreases, the perturbation on the location Z CD decreases, too.Thus, the search results to the optimal path selection in the search space, reducing the number of iteration.The optimal path selection of nectar bees is discarded by scout bees and it is replaced with new optimal path selection resulting in scouts clusters.In ABC, this is replicated by producing an optimal path selection location randomly and replacing it with the discarded path selection.In ABC, providing the best optimal path in cluster cannot be improved further during a predetermined number of cycles.The optimal path is assumed to be abandoned.Assuming that the discarded nodes paths is ˴ and ˪ ∈ {1, 2, . . ., ˖}, then the scout bee discovers a new optimal path to be replaced with ˴ .This operation can be defined as in (6): After each candidate optimal path selection it is estimated by artificial bee by pertaining to source position v CD is formed.Its accuracy is compared with the old path selection.If the new path selection results in an equal or higher than earlier path selection results, then current path selection in cluster is selected as best path, else the present path is kept as the best path.In cluster and is determined to be the best path to send packet from source to the destination.On the other hand the selected old path is kept in the memory.In other words, a greedy path selection is employed as the best path selection among old and new path.There are three major control parameters in the ABC algorithm to be considered.They are the number of optimal food sources which is equal to the Single population (SN), the value of maximum value (MCN (Maximum Cyclic Number)) to total number of the path selection.In a robust path selection process for each cluster, investigation and development processes are required to be carried out together in WSN.In the ABC algorithm, where both onlooker and employee bees carry out the investigation process in the path selection, the scouts bee control the development process of path selection in cluster.The local path selection results in ABC algorithm.The algorithm pertaining to the nearest node of cluster path investigation and greedy selection of each node is carried out and deployed in between the employed bee and onlooker.The entire path selection result in random search procedure performed by scouts'.Thereby the nearest path of the employed and onlooker bees is selected.
Proposed algorithm for optimal path selection using artificial bee colony: Input: Cluster head and cluster member Output: Best path discovery for cluster member to cluster head Generate initial population; each cluster is considered as single population ˴ , ˩ = 1, … ., ˟˚ SN is the cluster size Evaluate the fitness of each node in the cluster Set I˳Iˬ˥ = 1 Repeat For each employee bee { Produce a new optimal path with trust value ˰ using v CD = z CD + ϕ CD z CD − z ED Where z CD is a path between the node ˩ and ˪ Calculate the fitness value ˦ for new path Apply the greedy selection to select the best optimal path from the set of new paths } Calculate the probability value J for each optimal path selected from Eq. ( 4) For each onlooker bee { Select a best optimal path ˴ depending on probability J Produce new path solution ˰ Calculate the fitness value ˦ for new path Apply the greedy selection to select the best optimal path from the set of new paths If there is any abandoned solution for scout bee then replace it with new optimal path solution which will be randomly produced by ( 6)

EXPERIMENTAL RESULTS
The performance of the proposed improved Role trust model from following aspects: First to decrease network packet loss rate, less energy consumption and higher throughput.In the proposed model, there is no difference between client's nodes and server nodes.During experimentation the trust management system with role, uses a horizontal and rectangular area of 800×400 m.On observation the proposed work reduces packet drop ratio and increase throughput.The Table 1 shows the simulation for experimental results.
In this graph Fig. 2 shows the packet drop ratio between trust management system among ABC and improved Role Trust management system, it shows that proposed system have less drop ratio than without Role trust in ABC.
Figure 3 shows the energy consumption between trust management system among ABC and improved Role Trust management system, it shows that proposed system have less energy consumption than without Role trust in ABC.
Figure 4 shows the throughput between trust management system among ABC and improved Role Trust management system, it shows that proposed system have more throughput than without Role trust in ABC.

CONCLUSION
In this study introduce the Role based Trust Management System (RTRMS) framework for wireless sensor networks.In this study role based trust management system elects cluster head based on roles and remaining nodes in cluster are named as cluster members.In this study we have proposed an Artificial Bee colony Role Trust Management system for WSNs, called ABCRTRMS-WSN.It has been seen Shown that bee colony food sources deposited by onlooker bees help next bees to find the most trustworthy server through the most reputable path all over the network.The resulted ABCRTRMS-WSN is more efficient and appropriate for universal WSN with less packet drop ratio and high throughput.In particular, the ABCRTRMS-WSN is very successful in decreasing energy consumption and packet drop ratio with higher throughput ratio than without role based ABC model.It shows that ABCRTRMS-WSN framework is sufficient and has high accurateness in avoiding malicious nodes from being a cluster head selected based on role trust framework.

Fig. 1 :
Fig. 1: The proposed architecture for role trust based cluster formation and optimal path selection among the eligible nodes.The cluster head is selected based on the calculation of trust value, residual energy for each node and based on that calculation it forms roles such as Cluster Head (CH) and Cluster Member (CM).The calculation of the trust values and role formation are given below.Trust value calculation: Time based past communication trust evaluation: Node level trust value is estimated based on earlier time ST L,M = The time of UT L,M last successful past communication ∆ˠ = The threshold time