MoB: A Mobile Bazaar for Wide area Wireless Services R. Chakravorty, S. Agarwal, S. Banerjee, I. Pratt T 110.7190 Research Seminar on Telecom Software 28.09.2005 Yrjö Raivio 28916V T 110.7190 Research Seminar on telecom software/28.09.2005/yr 1
Motivation Mobiles lack internet connectivity WLAN: no coverage everywhere Cellular: expensive, high latency, low BW, occasional interruptions Cellular and service providers are coupled, not possible to select speech and data services from different operators Long term agreements T 110.7190 Research Seminar on telecom software/28.09.2005/yr 2
Mobile Bazaar (MoB) Design Targets Open Market Collaborative Architecture Improve Wide area wireless data service performance by 1) Decoupling infrastructure from service providers 2) Allows service interactions on arbitrary timescales 3) Flexible compositions of service interactions Multiple providers simultaneously Anybody can sell services, reselling Prof. Raymond Steele: Anybody should be able to make services on any access, 3G 2001 Competition T 110.7190 Research Seminar on telecom software/28.09.2005/yr 3
Device roles: Roles in the Service Business Web sites and Applications roles: Web Service Consumers Web Service Providers Mobile operator roles: Web Service Provider Web Service Consumer Service protocols Authentication Discovery Profile Location Charging Messaging Presence Web Service Consumers Web Service Providers T 110.7190 Research Seminar on telecom software/28.09.2005/yr 4
Service Value Chain Terminal Radio network Core network Network Application ISP Portals Content providers 1... 2 N 1 2... N 1 2... N 1 2... N 1 2... N 1 2... N 1 2... N T 110.7190 Research Seminar on telecom software/28.09.2005/yr 5
Introduction Service examples: Bandwidth, Location, Time, Web Proxy caching, Search, Filtering Hop by hop, no routing protocols needed Mainly applicable to proximity range utilizing WLAN or BT On Application layer Free market, no control Incentive, everybody can set their own prices Reputation management Billing and Accounting T 110.7190 Research Seminar on telecom software/28.09.2005/yr 6
Architecture Vito Components: 1) Internet connectivity 2) Device to device connectivity 3) 3rd party accounting, billing and reputation management Service trading decisions based on user preferences and policies, e.g. battery low Trust or no trust, Billing or no billing Based on ebay trust management system, accounting added T 110.7190 Research Seminar on telecom software/28.09.2005/yr 7
Operation (1/2) Each user registers with Vito to get a timestamped Reputation Certificate (RC); includes history information During the trade RCs are checked and new rating values are updated off line from and to RCs via Vito Feedback can be given by both the Trader and Customer Malicious Customer = Gives low scores independently of the service Malicious Trader = No service after payment Service first, pay next strategy Reputation risks Sybil attacks: malicious user acquires a new identity Collusions: group of users rate each other high Decentralized systems T 110.7190 Research Seminar on telecom software/28.09.2005/yr 8
Operation (2/2) T 110.7190 Research Seminar on telecom software/28.09.2005/yr 9
Experiment File Transfer T 110.7190 Research Seminar on telecom software/28.09.2005/yr 10
Simulation Location Accuracy T 110.7190 Research Seminar on telecom software/28.09.2005/yr 11
Good vs. Malicious Users T 110.7190 Research Seminar on telecom software/28.09.2005/yr 12
Basic idea is good but Concluding remarks Idealistic P2P model copied to mobile world as such A lot of similar academic research has been carried over with minor practical success Vito looks pretty nice solution but in a large system scalability can be questioned Malicious traders are not a problem due to reputation management Experiment and Simulation models look suspicious In general paper could be more compact and scientific T 110.7190 Research Seminar on telecom software/28.09.2005/yr 13
Challenges Mobile Operators do not like this; where is the incentive? Reselling can violate copyright How to prevent that users start to distribute free of charge? DRM? Hop by hop, what about the End to End security? Might work in Proximity but current data rates prevent the usage of cellular networks System does not react online if Customer exceeds his/her credit Would be interesting to see simulation on selection policy where Traders with different accesses (= prices and performance) were compared Discussion! T 110.7190 Research Seminar on telecom software/28.09.2005/yr 14