Data Management Middleware Protocols In Ad-hoc Social Networks

Ad-hoc social networks (ASNETs) provide infrastructure-less settings for mobile users to communicate with each other opportunistically. Taking advantage of the users’social charac-teristics collected from modern sensor devices can enhance and fine tune the performance of mobile ad-hoc networking. Middleware design for this socially-aware paradigm is of paramount importance to develop innovative applications and services easily and efficiently. Nevertheless, due to various limitations such as lack of centralized management, device heterogeneity, unre-liable wireless communication, mobility, resource constraints, or the need to support different traffic types, a number of new challenges have emerged recently. One of the main challenges is performance degradation and implementation inefficiencies, which are caused by the gap be-tween the application and lower layer protocols. In addition, managing data poses a severe challenge for both human beings and wireless networking systems, which usually only model protocols by ignoring or within a certain consideration of users’social and mobility informa-tion. ASNETs can be of immense use in case of pervasive conference/meeting and ubiquitous emergency situations. Most scenarios of this network clearly illustrate the relevance of incorpo-rating important services or modules (e.g., data management) with middleware design for proper wireless network operation.In previous works, ad-hoc networking frameworks have been proposed and data manage-ment issues were investigated in which social awareness was not considered. To address the downside, we focus on developing a formal framework for ASNETs middleware upon which de-sign and proposal of our data management protocols can be based. We take advantages of social networks and mobile ad-hoc networks to integrate social-awareness and user mobility, respec-tively. This leads us to investigate various aspects of data management middleware problems including data availability, load distribution and users’cooperative participation in partitioned communities. ASNETs differ from social networks in which disconnections are the norm in-stead of the exception. Traditionally, different ideal approaches are widely used to facilitate data availability, even load distribution and users cooperation. However, the quality of wireless links would be affected by many factors like mobility, overhead and users’selfish behavior. A data replication methodology and community-partitioning concept is applied intensively throughout the course of this study to tackle the aforementioned challenges. To the best of our knowledge, this is the first work to explore data availability, load distribution and selfishness issues together in ASNETs.The accessibility and availability of ASNET services can be assured by replication ap-proaches. Replica allocation helps to avoid data losses in case of an unpredictable group mobil- ity that causes community partition and also aids in reducing the number of hops when a data is transmitted from source to destination. However, it is impossible to replicate all data items on every node because of the limited resources. The question here is, where and how we should allocate the data replicas to gain high data accessibility and availability. We first contribute to this line of research with replication protocol for maximizing availability of a data by proposing ComPAS, a community-partitioning aware replica allocation method. In ComPAS, we apply an efficient and consistent way to store replicas for each user’s data. Consequently, we choose the numbers of replicas depending on the replication budget of the system and its desired avail-ability. Its goal include integration of social relationship for placing copy of the data in the community to achieve better consistency by keeping the replica read cost, relocation cost and traffic as low as possible. It tries to balance the community storage space load as next goal. ComPAS offers a more interesting pattern in terms of read cost as compared to other schemes and it is highly efficient when relocation of replicas happened in the network.ASNET systems are poised with challenges of performance degradation and poor scalabil-ity. This is typically caused by an uneven load distribution of operations and the susceptibility of link failure. The fair functionality of any distributed system can be realized by allowing its integral computational elements to work cooperatively. An effective load balancing mechanism guarantees optimal use of the system resources whereby no broker remains in an under-loaded state while any other broker is being overloaded. In many of today’s distributed environments including ASNETs, users are linked with limited resources such as storage space and bandwidth that inherently inflicts tangible delays. This leads the users to rely on inter-resource communica-tions and load exchange throughout the network. To be able to fully benefit from such network-ing systems, data availability and resource distribution are key services, where issues of load distribution, partitioning and fault tolerance present a common challenge. Herein, we enhance the data management performance by proposing Co-Lab, community-based event dissemina-tion with load balancing and fairness mechanism. This protocol employs interest similarity and filter replication approaches for clustering brokers in a community. Its goals are to achieve bet-ter load distribution more uniformly among brokers and circumvent highly overloaded brokers by keeping the reliability as high as possible. Performance evaluations indicate that Co-Lab has promising advantages by achieving relatively better load balance, reduced overall load, and robustness against failures.Another shortcoming of existing data management protocols is the assumption that users are cooperative when participating in operations such as forwarding data. If nodes refuse to collaborate in the network services, end-to-end connection may not be possible. Such unwill-ing (selfish) behavior can greatly degrade the network performance. Therefore, it is essential to detect such selfish users and mitigate their impact on the performance of other well-behaving users. Even though solutions for detecting selfish users have been explored before, a few funda-mental shortcomings surrounding the problem have remained unaddressed, particularly in data dissemination and forwarding for ASNET services. Thus, we attempt to design an algorithm that gives each user the autonomy to identify and exclude selfish users. Features like these can often be observed in biological processes such as bacteria and inspired us for augment-ing our new cooperative architectural concept to the replica allocation protocol. In this work, we propose a biologically inspired algorithm to detect and mitigate the impact of selfish users called BoDMaS that employs ComPAS as a data replication model. Using social and biological mechanisms, BoDMaS assesses and classifies users, and denies selfish user participation. The effectiveness of the proposed scheme is evaluated using different metrics selected for evaluation, demonstrating its ability of accurately detecting selfishness in replication operations for ASNET environments.