Distributed Data Retrieval for Real-Time Decision-Making under Freshness Constraints

Feng, Shuo

Permalink

https://hdl.handle.net/2142/88910 Copy

Description

Title

Distributed Data Retrieval for Real-Time Decision-Making under Freshness Constraints

Author(s)

Feng, Shuo

Contributor(s)

Abdelzaher, Tarek F.

Issue Date

2015-12

Keyword(s)

• distributed systems
• crowd-sensing

Abstract

This paper describes a distributed data retrieval algorithm for crowd-sensing application, which aims to collect data with minimized bandwidth cost while satisfying data freshness constraints. In a resource-limited setting, data loses freshness very fast. For instance, the condition of a road during a rush hour may be dynamic due to the rapid change of the traffic. In order to schedule an optimized route to a destination from a given location, we have to know its real-time condition. The protocol we design is to exploit logic dependencies among data by using and-or tree to reduce the overhead of the network and handle concurrent requests at the same time. Meanwhile, we further modify the centralized system into a distributed form so that each node in this network is able to calculate the best retrieval order locally. Furthermore, we integrate some ideas of other literature to let each node store the retrieved data locally due to the fact that the price of storage is lower and lower these days. Finally, we implement part of the algorithms and test the efficiency of using varying probabilities and earliest deadline first to sort queries.

Type of Resource

text

Language

en

Permalink

http://hdl.handle.net/2142/88910

Owning Collections