*
Quick Links|Home|Worldwide
Microsoft*
Search for


Narasimhan Nitya

Ad Hoc Social Networking: Where proximity meets affinity

Contact Information
1301 E Algonquin Road, IL02-2240
Schaumburg, IL 60173

Biography
Nitya Narasimhan is a researcher in the Pervasive Platforms and Architectures Lab at Motorola, with current research focus on distributed application architectures, service middleware for mobile ad hoc networks, near field communication and peer-to-peer platforms. Her doctoral research focused on reliability for distributed Java systems.

Position Paper
Social networking leverages established relationships between some users to enable the creation of new relationships between many others. Internet-based sites such as LinkedIn, Friendster, and Tribe.Net have leveraged such networks to provide services such as job-finding, dating and recommendations. These centralized, infrastructure-based networks benefit from the size and diversity of the member population but can be influenced by the vagaries of publicly-articulated identity—in particular, such networks are driven by their implicit trust in the user-specified relationships and can therefore be undermined by poor status updates or by the limited options in establishing the degrees of relationship beyond a binary “yes or no” (friend or stranger).

By comparison, an ad hoc peer-to-peer network limits its users to the more opportunistic interactions that occur only when they are in the physical proximity of other members. A social network created by such serendipitous encounters is likely to be relatively smaller in size and reach. However, because such interactions are spontaneous (require minimal effort from the user), they are also likely to provide an unbiased and dynamic view of the status of the user’s relationship with peers in that social network. In other words, the duration, purpose and frequency of encounters can be used to determine automatically the potential for a social link between the participants and to establish this link after ratification by the user. Furthermore, automated tracking of any subsequent interactions between the users can be exploited to establish the degree of the relationship (friends and family, colleagues, peer acquaintances and familiar strangers ) and maintain updates on its validity over time.

With the Acapella middleware platform, we have begun exploring the utility and challenges of supporting such p2p services over ad hoc wireless LANs. Acapella is mobile Java middleware that provides APIs for enabling three classes of behavior—federation, tribal and encounter—that underscore the “Share-Interact-Notify” capabilities required for social interaction. Each requires or involves a different degree of trust and familiarity between users. Federation focuses on remote access to shared data; typically, such services are extended only to trusted peers. Tribal focuses on interactive services such as gaming or chat; such services are non-invasive and can be extended also to “suitable” strangers. Encounters focus on alerting the user to specific events of interest; encounters are asynchronous and can be activated both by known peers or by strangers that match specific criteria.

In the presentation, we will focus on the challenges (energy-efficiency, user-transparency and reduced complexity) faced in developing the Acapella platform and discuss insights gained from a user study conducted using both storyboards and working prototypes. The goal is to motivate a discussion on how such ad hoc p2p platforms can both complement and validate infrastructure-based social networks, thus increasing their value in enterprise and consumer environments. This will require rethinking the definition of social identity (to enable its seamless translation and tracking across both infrastructured and ad hoc environments) and addressing related issues such as privacy and access control. Finally, it will enable social networking “routes” to be evaluated based not only on hop-distance, but also on most-recent activity.

 

Back to Social Computing Symposium 2005

 


©2008 Microsoft Corporation. All rights reserved. Terms of Use |Trademarks |Privacy Statement