Regression based critical information aggregation and dissemination in VANETs: A cognitive agent approach

Abstract

Data aggregation in Vehicular Ad hoc NETworks (VANETs) is an efficient technique for effective usage of communication resources. In dense VANET traffic scenarios, data aggregation is needed to represent several numbers of almost similar critical information into one refined critical information to reduce bandwidth requirements. This paper proposes a cognitive agent based critical information aggregation and dissemination in VANETs by using regression mechanism. Regression based cognitive agent approach efficiently aggregates the collected critical information and minimizes redundant data dissemination. The proposed scheme works over clustered vehicles by using a set of static and mobile agents. The scheme operates in the following steps: (1) validation and filtering of collected critical information; (2) generation of beliefs based on valid and filtered critical information; (3) aggregating the beliefs to develop desire using a regression technique; (4) revision of desire for better quality of aggregation; (5) finalizing the intention based on revised desire; (6) disseminating aggregated information to neighboring clusters. We validate the proposed scheme by simulation. The scheme performs better as compared to ESSMD (Event Suppression for Safety Message Dissemination) scheme in terms of critical information acquisition delay, aggregation delay, end-to-end delay, dissemination delay and bandwidth utilization.

Introduction

Vehicular Ad hoc NETwork (VANET) is an example of Mobile Ad hoc NETwork (MANET), where mobile nodes are vehicles. Moving vehicles equipped with communication devices form exactly an instance of long envisioned MANETs. Communication is possible between vehicles within each other's radio range as well as with fixed road side infrastructure components. VANET is an integral part of the intelligent transportation system (ITS) architecture which helps to improve the road safety and optimize the traffic. Traffic accidents resulting in injury and death as well as traffic congestion are caused by ever increasing number of vehicles on the roads. This is mainly due to lack of safety services information. Traffic congestion results in heavy delays for the travelers and creates high emission of substances harmful to the environment. Hence critical information related issues are given highest priority in VANETs [1], [2], [3], [4], [5].

Creating high speed, highly scalable and secure VANETs presents an extraordinary task due to a combination of highly dynamic mobility patterns and network topologies. VANET raises several interesting issues with regard to media access control (MAC), information gathering, information dissemination, information aggregation, information validation, routing, network congestion, performance analysis, privacy and security [6].

Critical information data aggregation and dissemination with minimum delay is a challenging issue [7]. Limitations of existing aggregation schemes are as follows: lack of intelligence in aggregation, less flexible for dynamic varying critical information parameters, lack of combination of aggregation and dissemination mechanisms, low scalability and moderate data aggregation time. This paper proposes a cognitive agent based regression model for critical information aggregation and dissemination using clustering concept presented in [8] to overcome the limitations of existing work.

Cognitive agents are a class of software agents which are intelligent autonomous programs activated on an agent platform of either a host or network. These agents use their own symbolic representation knowledge and mentalistic notions based belief base to achieve specified goals without disturbing activities of a host. Mobile software agents are flexible modular entities which can be created, migrated, deployed and deleted in real-time. Mobile code should be platform independent, so that, it can execute at any remote host in a heterogeneous network environment [11], [12], [13]. Usage of mobile agents for development of new applications in vehicular networks is discussed in [14].

The properties of cognitive agent are as follows: carry out activities in a flexible and intelligent manner, rapid response to environment changes, learn from experience, communicative and co-operative with other agents and proactive, i.e., exhibits opportunistic, goal-oriented behavior and takes the initiative appropriately. A framework of reflective–cognitive agent architecture which enables agent to alter its own code in run time according to the changes in environment is proposed in [15]. Some real implementation of cognitive agents are as follows. Cognitive agent implementation in the Blender Game Engine (BGE) environment is proposed in [16]. A cognitive multi-agents architecture for designing intelligent and adaptive learning systems is discussed in [17].

Belief–Desire–Intention (BDI) model of cognitive agents has been widely used in dynamic and complex scenarios where incomplete information about the environment and other agents is available. The BDI model provides an explicit declarative representation of three key mental structures of an agent: informational attitudes about the world (beliefs), motivational attitudes on what to do (desires) and deliberative commitments to act (intentions). BDI agent reaction for incoming event sequences for time-sensitive applications like the Close-In weapon system in air-carriers is described in [18]. Estimation of average response time using the average attributes of a sequence of events based on probability and queuing theory is performed. BDI model has become predominant architecture for the design of cognitive agents [19]. BDI modeling of pedestrian behavior in a real-world environment is discussed in [20].

A BDI-based framework for a cognitive agent that acts as an assistant to a human user by performing tasks on user behalf is given in [21]. Rapport–Belief–Desire–Intention–Adaptation (RBDIA), a method to support progress from individual autonomous agent concept towards a collaborative multiple agents is discussed in [22]. Some of the benefits of BDI architecture for aggregation in VANETs are as follows: (1) BDI model has the capability of quick adaptivity and learning in VANET environment. (2) As critical information in vehicle environment is continuously changing, beliefs about environment can be updated regularly in BDI architecture. (3) Autonomous decision on aggregation can be taken on available beliefs related to critical information. (4) BDI model creates commitments and performs action on the basis of intentions with full commitment, and (5) BDI model provides an explicit model of teamwork which is critically required in VANETs.

The clustering concept used in the proposed work is as follows. Multiagent driven clusters of vehicles are formed in VANETs at lane intersections by considering vehicle speed, direction, connectivity degree to other vehicles and mobility pattern. Cluster members are identified based on vehicle's relative speed and direction. Cluster head is selected based on stability metric derived from connectivity degree, average speed, and time to leave the road intersection. Cluster head predicts future association of cluster members based on mobility patterns. The announcement of cluster mobility pattern to all cluster members is made by cluster head. The cluster members with similar mobility pattern can reconnect with cluster head after passing an intersection of the lane.

Each vehicle is equipped with a cognitive agency that comprises of static and mobile agents. The agents used in the proposed model are Vehicle Manager Agent (VMA), Critical Information Collecting Agent (CICA) and Aggregated Information Dissemination Agent (AIDA). The agents are located in cognitive agency of a vehicle. VMA is a static cognitive agent whereas CICA and AIDA are mobile agents. VMA uses Belief–Desire–Intention (BDI) model to employ regression based cognition for critical information aggregation.

The proposed scheme operates in the following steps after cluster formation. (1) VMA of a cluster head periodically triggers CICA to collect critical information in a cluster. (2) VMA performs the validation process by using validation scheme given in [9], and eliminates duplicates in collected critical information. (4) VMA generates beliefs based on the finalized critical information. (5) Beliefs are aggregated to generate desire using regression technique. (6) Desire is revised to achieve quality of aggregation. (7) VMA finalizes intention about the revised aggregated critical information, and (8) VMA triggers AIDA to relay vehicle of cluster to disseminate aggregated information.

Our contributions in comparison to existing works are as follows. (1) Design of cognitive agent based dynamic critical information aggregation and dissemination scheme to optimize overall delay. (2) Mobile agents usage for critical information collection reduces the network load, and (3) mobile agent based data dissemination approach helps to adapt in varying VANET topologies. The proposed scheme has been compared for its performance with Event Suppression for Safety Message Dissemination (ESSMD) scheme [10].

Rest of the paper is organized as follows. Related works are discussed in Section 2. Regression based critical information aggregation and dissemination using cognitive agents is given in Section 3. Simulation model for proposed scheme is presented in Section 4. Result analysis is given in Section 5. Finally, Section 6 concludes the work.