NOSQL Query Processing System for Wireless Ad-hoc

Sensor Networks Name -Bhaware Nikhil Uttamrao

Roll no- 137503

Sensors

Sensors are micro-electro-mechanical devices • Capability of sensing• Data processing capabilities• Communication capabilities• Low power devices

Wireless Sensor Networks

• spatially distributed autonomous sensors to monitor physical or environmental conditions.

Query processing systems for WSNs

• Provides abstractions for SQL query interfaces• Represents sensors on sensor networks as a table• User inserts query at base station and it converts those queries into

sensor node understandable format• Based on Relational database management system• Some popular query processing system 1)TinyDB(for TinyOS) 2)TikiriDB(for Contiki)

TinyDB

• Declarative database abstraction for TinyOS operating systems.

• Provides user with database file called sensor.• Uses resource aware algorithm to collect data• Acquisitional query processing• Supports data logging and network health monitoring• Queries can be prioritized

Cougar• Supports a platform for testing query processing techniques• It has three tier architecture

1)Query proxy2)Front end components3)Graphical user interface

• Designed for in-network processing

TikiriDB

• Database abstraction layer for contiki operating system• Provides sql query interface called TikiriSQL

ex. Select TEMP,HUMID from sensors SAMPLE PERIOD 2 FOR 10

Client with TikiriSQL Library

• Functions to issue SQL queries by user program• Main tasks

1)accept queries from user program2)parse the query and put it in managabale format3)if errors , returns warning to user

Need of NOSQL abstraction

• Working with large sets of data using table based database system needs lots of resources

• ACID properties does cannot be guaranteed in a sensor networks

Design of NOSQL database abstraction

1)NOSQL Query

• Design Query syntaxes based similar to NOSQL database which we will use for abstraction.

• ANTLR TOOL

ex. GET temp SET 2 FOR 100 ZRANK temp 2.0

2) Lexical Analyser and Parser

• i/p – NOSQL query• Lexical analyser generates tokens• Parser checks for syntax and semantics and generates

parse tree• Parser produces c code according to NOSQL query

3)Query Processor

• Processes c file generated by the parser• Distinguishes the parts of the query• Generates query message header and query payload

4)Data Packet Implementation

5)Serial Forwarder

• Assigns unique id to each packet• Stores query id and client id in its memory• Pass the packet in the network

6)Mesh routing

• Communication is done by using ad-hoc mode• Uses hybrid routing protocols

7)Redis Architecture

• Open Source• Uses key –value for storage• Various data structures such as strings , hash, list, sorted

list• Runs on port 6379• Implement RedisDB plug-in with Strings and Sorted Set

Performance Analysis on Cooja SimulationPlatform

• For TikiriDB database abstraction

SELECT temp FROM sensors SAMPLE PERIOD FOR 10

• For NOSQL database abstraction

•GET temp SET temp 2 FOR 10

Conclusion and Future Work

• NOSQL queries perform better than RDBMs Queries• High scalability• Query optimization

Thank You