Overview Issues in Mobile Databases – Data management – Transaction management Mobile Databases and Information Retrieval
Distributed system with mobile connectivity Full database system capability Complete spatial mobility Wired and wireless communication capability
Traffic control Taxi dispatch Emergencies services: police, medical, fire, … Car navigation Location-aware search and recommendation Social applications: information / opinion sharing and search among community
Limited wireless bandwidth and communication speed Limited power (battery life) Limited screen size for output Limited query capability for input Limited computing & storage capacity Less secure
Anywhere: can physically move around without affecting data availability Anytime: can store and retrieve any data whenever mobile connection is available Efficiency: can process mobile data efficiently Effectiveness: can give users what they really want (, and skip the less relevant…)
To build a ubiquitous information processing system given the inherent limitations of mobile devices and wireless communications
Data management – Data caching – Representation and storage of mobile data – Data classification Transaction management – Query processing – Concurrency control – Transaction and error recovery
How to improve data availability to user queries in mobile database systems? – Data caching Results of previous queries are cached on mobile clients Query log can be further used in personalization and recommendation – Representation and storage of mobile data Spatial data management Indexing and retrieval schemes – Data classification Location-dependent Location-independent
Results of previous queries are cached on mobile clients – Cache size – Cache freshness/update frequency – Cache consistency Query log can be further used in personalization and recommendation – User preference learning keywords, categories, location – Collaborative filtering for content recommendation
Spatial data management – Location information – Data volume, update frequency, persistency Indexing and retrieval schemes – High-dimensional feature space – Structure for improving retrieval efficiency
Location-dependent – Location-based services E.g. nearby gas stations or restaurants within 1km, available paths to the park, … – Database distribution or replication must take location into consideration Distributed vs. centralized Location-independent – Personal name, account information, plate number, …
Query processing Concurrency control Transaction and error recovery
Query types – Location dependent query – Location aware query – Location independent query
A query whose result depends on the geographical location of the query origin – E.g. What are the 3 nearest gas stations or restaurants? What is the shortest path to the park? – GPS can facilitate this Outdoor only WLAN could cover most of the indoor locality
A query whose context might be related to locations – E.g. Find out all the car plates passing through the traffic light between 7:00-8:00pm Find out the current location of the bus on route #212 Find out the top 5 roads that have the most traffic in Taipei (the slowest car speed)
Similar to the issues of concurrency control in distributed systems – Time synchronization Timestamps, clocks – Latency in mobile queries – Similar issues to cache consistency Size of query results
Conventional transaction properties – Atomicity – Consistency – Isolation – Durability Too rigid for mobile database – Flexibility can be introduced
Transactions – Wireless communication availability and overhead – Hard to manage locking and unlocking operations – Limited power Recovery – Efficient logging and check pointing – Log duplication
In mobile e-commerce, security in mobile data and transactions are critical – Less support in wireless security WEP, IEEE i – Identity authentication – Privacy issues of user location
Indexing and storage of mobile data Efficient query processing and retrieval of mobile data Content adaptation and information presentation for small display Automatic summarization and personalization of mobile data Scalable algorithms for large mobile databases Knowledge sharing among mobile peers Security and trusted retrieval of mobile data Location-based search for mobile devices
Mobile information retrieval – Context awareness – Content adaptation Context Awareness Location information Built-in camera Social networks
Location-based search Spatial data mining Query log mining User profiling and recommendation Knowledge sharing
Small screens Less processing power Less memory and storage
Automatic summarization Information filtering User personalization Efficient indexing and retrieval of mobile media Scalable algorithms