This paper proposes a novel middleware for service selection in mobile ad-hoc networks with a particular focus on scenarios immediately and subsequently afterwards an emergency. The proposed middleware, operating on a mobile user's hand-held device and collecting the user's contexts through machine-to-machine connectivity, has three major contributions as compared to the current literature. The middleware, based on the collected contexts, firstly classifies a service request, e.g., safety-related or comfort service. Then, it initiates the required network connectivity for service discovery, i.e., the ad-hoc connectivity when an infrastructure-based network is inaccessible. Finally, the middleware selects the service based on a method specifically proposed for that service category, including pre-defined realistic user contexts, to access it. The simulation results show that the middleware achieves up to 12 percent higher success rate compared to the minimum hop count service selection method. Compared with the same method, it has a response time up to 50 percent lower for the safety-related but 80 percent higher for the comfort services. Yet, the middleware attains up to 96 percent user satisfaction rate for both service classes.

A global trend toward large scale emergencies has placed an emphasis on the achievement of a ubiquitous public safety network. Such a network may be realized over a mobile ad hoc network formed by the handheld mobile devices. Therefore, traffic in the network can be user generated and thus semantic. Unfortunately, none of the traffic management techniques proposed for the underlying network considers the semantic properties of the generated traffic. Therefore, in this paper, we propose a semantic traffic management framework which has two modules: traffic monitoring unit and traffic reduction unit. Although the first module analyzes the semantic traffic to detect an emergency, the latter module removes redundant semantic information for traffic reduction. We have supported the feasibility of the proposed semantic framework through simulation. Simulation results suggest that the framework is capable of accurate and early detection of an emergency as well as traffic reduction while keeping sufficient information to report the emergency. Copyright © 2012 John Wiley & Sons, Ltd.

The number of emergencies has been observed to increase exponentially in recent years. Mobile devices, being ubiquitously present with users, often found to be a preferred way of communication in these emergencies. Since realistic prediction of user movements in emergencies is still an open issue, recently a machine to machine (M2M)-based movement prediction framework has been proposed. The proposed framework considers user's behavioral changes as well as simple geographical constraints initiated by emergencies. Nevertheless, these considerations are not sufficient to present a practical emergency environment and thus realistically predict emergency-affected user movements. Consequently, this paper proposes a modified M2M-based framework for movement prediction. The modified framework considers not only user's behavioral changes but also following factors: a realistic pre-emergency mobility model, arbitrarily-shaped obstacles, a signal propagation model and a practical personal belief computation. Simulation results have been used to verify effectiveness of the proposed framework.

The ever-growing popularity of handheld mobile devices has been accelerated by their ubiquitous presence to support user mobility. In mobile environments, a prior prediction of user movements can improve both network and application-level performances. In the existing literature, a major research on mobility has focused on user behavior to predict their movements. Although emergency-affected users are more likely to deviate from their usual behavior, the influence of an emergency on user behavior and thus user movements has not been investigated until now. However, an accurate user movement prediction in emergencies is often more important due to post-emergency resource constrains in the network. Being closely carried by their users, mobile devices can be conveniently employed to monitor user behavior in emergencies and consequently predict user movements. Therefore, this work proposes a movement prediction framework for mobile users in emergencies using a machine-to-machine (M2M) network of mobile devices. Simulation results exhibit that the prediction accuracy reaches up to 100%, i.e, predicted movements by the proposed framework are exact to the real movements.

Of late, an easy access to the Web 2.0 applications through the hand-held mobile devices has provided the consumers with the opportunity to express opinions on their invoked services conveniently. These opinions, also known as the Internet word of mouth, often impact a potential consumer's service selection notably in service rich urban terrains. Access to these opinions involves tariffs since they are available over the Internet. In addition, individual analysis of such opinions is inconvenient due to mobile devices' inherent limitations. However, a `pervasive cooperative consumer network' can be employed using consumers' connectivity enabled mobile devices to share and acquire service related opinions at no monetary cost. Yet, this network, realized over an underlying mobile ad hoc network, does not consider inclusion of the opinions for service selection. Therefore, we propose a service selection framework to integrate opinions with the widely used minimum hop count method for service selection. The opinions are first clustered using the Wordnet-based K-means clustering algorithm and then quantified using the Sentiwordnet scores to be automatically included in the framework. The proposal is further supported by the simulation results.

The continuing trend of large scale emergencies in last decades have emphasized on ubiquitous public safety services. The growth of mobile handheld devices have provided people with wide scale flexibility of on demand services and enjoyment of ubiquity. In addition, currently people are not only service consumers but also information providers to realize ubiquitous public safety networks. During and right after an emergency, people seek safety and more likely to explode the underlying mobile ad hoc network with numerous reports on the emergency. These human originated reports may not be syntactically identical so the traditional traffic management schemes are not adequate to handle such explosion. Yet, these reports are either semantically or conceptually similar. Therefore, we propose a semantic agglomerative traffic management framework to efficiently remove the aforementioned semantic information redundancy by employing the Agglomerative Clustering and Latent Semantic Analysis techniques. Simulation results are presented to further establish the worthiness of the proposed framework.