A distributed architecture to support infomobility services Claudia - - PowerPoint PPT Presentation
A distributed architecture to support infomobility services Claudia Canali Riccardo Lancellotti University of Modena and Reggio Emilia Motivation Web 2.0 Web 1.0 Dynamic Web-based Static Web pages services Information
2
– Static Web pages – Information repository – Limited interactivity
– Static maps – Basic navigation support – No interactivity – Car-oriented services
– Dynamic Web-based
– Personalized services – Collaborative services – High interactivity
– Collaborative infomobility
– Personalized services – High interactivity (wireless
– Services oriented to every
– Real-time POI sharing – Geo-referenced bulletin boards – Maps updated by the users for the users (Wikimaps)
– Automatic detection of delays, traffic jams depending on
– The infomobilty system does not rely only on external data
– Real-time POI sharing – Geo-referenced bulletin boards – Maps updated by the users for the users (Wikimaps)
– Automatic detection of delays, traffic jams depending on
– The infomobilty system does not rely only on external data
– User data privacy – Data consistency – Service performance – Service availability
Geographic data, maps Support Info User Info
– preliminary experiments with Web
– central node, first mile, DoS attacks
– Function replication
GIS data Support DB Maps and geographic data for a cell
User DB
– Critical information only on central system – Use of temporarily IDs for interaction with edge nodes
– Data partition on the edge nodes and on central system
– Central system → clustering – Edge nodes → replicated
– Most interaction is with edge nodes
– Apache httpd + Tomcat – Axis 2 as the Web service implementation – GRASS GIS – Mysql
– 1 Apache httpd dispatcher – 4 Tomcat + Axis2 + GRASS
– Tomcat + Axis2 + Mysql
– Route ID – Expiry date – Cell for which the token is valid – User info: reputation,... – Signature of the central system
– Maps – POIs
– New POIs – Information with global relevance (e.g., public
– User position and speed
Warning to central system Warning DB Traffic Jam detection User Speed aggregation Speed limits Road characteristics Mobile users by car Train delay detection Train expected position User postion aggregation Train schedule Mobile users by train User speed User position
– Compromise between fully distributed and centralized
– No bottlenecks – No single points of failure
– High security – Access control,
– No bottlenecks – No single points of failure
– High security – Access control, frequent security audit
– Privacy → OK – Performance and scalability → Possible bottlenecks – Availability → Single Point of failure
– Performance and scalability → OK – Availability → OK
– Privacy → High security in every node is expensive
– Calculation of the user route – Authentication of the users – Accounting (pay-per-user services) – Generation of auth tokens for the interaction with edge servers – Access to external data sources (e.g., traffic status,
– Geographic data for the computation of user routes (GIS) – User preferences – Additional databases (e.g. public transportation schedules)
– Servicing user request for geographic data (e.g., nearby
– Updating geographic data based on user-supplied
– Extraction of information from user behavior – Aggregation and notification to central system of
– Maps, POIs, speed limits and other Geographic data about
– Additional databases (e.g. public transportation schedules)
– High number of edge nodes – Central system only for few, critical operations
– High number of edge nodes – Data replication allows an edge node to “take over”
– Central system is secure – User-related information are stored only on the central
– Use of temporarily ID for user interaction with edge