Name : Dinesh Bilimoria Date: Sept 25 th , 2013

1. CSLI5350G - Pervasive and Mobile ComputingWeek 2 - Paper Presentation“Multi-scale query processing in vehicular networks” Name: Dinesh BilimoriaDate: Sept 25th, 2013

2. Research Paper Bibliography: T. Delot, S. Ilarri, M. Thilliez, G. Vargas-Solar, and S. Lecomte. {2011) “Multi-scale query processing in vehicular networks”. Journal of Ambient Intelligence and Humanized Computing, 2:213–226.

3. Background What is a Mobile Query? • A request for information from a mobile device which can be either a push-based or pull-based query. What are some of the types of Mobile Query? • Location DependentQueries (LDQ) - queries whose answers depend on the locations of certain moving objects. • SpatiotemporalQueries (STQ) - include all queries that combine space and time and generally deal with moving objects. • ContinuousQueries (CQ) - whose answer is automatically refreshed as needed, in order to support the frequent changes/updates in the query results. What is Multi-Scale Query Processing? • Multi-scale query processing is any query processing that may need to access data sources of different types to obtain the answer to a query.

4. Multi-scale query processing in vehicular networks Objective • The goal is to provide query processing in a high dynamic vehicular networks. Previous research relied on push model. • Exploit different access modes (e.g., push, pull) and various data sources (e.g., data cached locally, data stored by vehicles nearby, remote Web services, etc.) to provide the users with results for their queries. Proposal • Focus on the specific case of vehicular networks where vehicles can exchange data among themselves to inform drivers about interesting events (e.g., available parking spaces, accidents, traffic congestions, etc.)

5. Multi-scale query processing in vehicular networks Example • Retrieve the list of petrol stations located in a radius of 10 km around me where fuel prices are less than 1$ (and update the result every 5 min). Fuel Price = 90c Fuel Price = 92c radius of 10 km Data is coming from different sources Fuel Price = 1.01c

6. Multi-scale query processing in vehicular networks Contributions • Problems are seen in V2V communications due to the high mobility of the vehicles and the unreliability of the wireless communications in such a dynamic environment. • Not to focus on one particular access model but rather to consider multi-scale query processing, which implies exploiting the available data sources whatever the access mode (e.g., push or pull). Assumptions • Open world assumption, as opposed to closed world. • Availability of multiple data sources • Retrieve the maximum number of results interesting for the user with respect to one or more criteria (e.g., result computation time, energy spent, financial cost, etc.)

7. Multi-scale query processing in vehicular networks Representation of Data Sources • Represent the properties of each data source by storing information about the different data sources accessible by the mobile device in an XML file • Necessary to have precise information about the I/O parameters for each data source • Decompose the query into several local queries to execute on the data sources • The different data sources can be local or remote data sources • For eg A local positioning service which provides to the user her/his location. Two different data sources providing the location of the mobile device ie (1) a GPS interface and (2) a WiFi based positioning service • An XML file stored locally on the mobile device • The XML element connector allows to define how the data source can be used.

8. Multi-scale query processing in vehicular networks Script Description of Data Sources

9. Multi-scale query processing in vehicular networks Optimization of Multi-Scale Mobile Queries Generation of all Possible Candidate Queries Attributes for Optimization 1. Accuracy 2. Average Availability 3. Cost – based on time, money and energy 4. Data Production Rate 5. Response Time 6. Update Frequency Cost of Query for each dimension Global Cost of Query where the goal is to minimize C(Q) for Optimization

10. Multi-scale query processing in vehicular networks Evidence • Cost of a query can be based on time, money and energy which was proved using a multi-scale query processor prototype. Evaluation of Prototype • Developed a prototype using Microsoft Language-Integrated Query (LINQ) API to evaluate multi-scale query processing • Benefits of using LINQis the possibility to query different types of data sources (a data structure, a Web service, a file system, or a database) • Developed two types of external LINQproviders; (1) translate a LINQ query into an equivalent HTTP GET request and (2) transform a LINQ query to a specific Web service using SOAP Cost query optimizer selects Candidate Query 2 (CQ2) since cost is minimized

11. Multi-scale query processing in vehicular networks Shoulder of Giants • This research was build on previous study using VESPA – Vehicular Event Sharing with a mobile P2PArchitecture. Impact • Cited by a new research paper on context-aware routing vital for intelligent inter-vehicular communication (2012) VESPA Prototype

12. Multi-scale query processing in vehicular networks Open problems • New data management, dissemination and query processing techniques are required to analyze the efficiency of context-based collaboration in VANETs. • Also to resolve query and result routing problems Discussion points • What is a Mobile Query?A request for information from a mobile device. • What is Multi-Scale Query Processing?Any query processing that may need to access data sources of different types. • What are some of the different types of data sources? A data structure, a Web service, a file system, or a database.