providing end-to-end delay guarantees for multi-hop wireless sensor networks

16
Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks I-Hong Hou

Upload: cecile

Post on 16-Feb-2016

37 views

Category:

Documents


0 download

DESCRIPTION

Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks. I-Hong Hou. Motivation. Wireless sensor networks are being deployed for real-time surveillance. Challenges. Wireless sensor networks can be deployed over a large area - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Providing End-to-End Delay Guarantees forMulti-hop Wireless Sensor Networks

I-Hong Hou

Page 2: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Motivation

• Wireless sensor networks are being deployed for real-time surveillance

Page 3: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Challenges

• Wireless sensor networks can be deployed over a large area

• Multi-hop transmissions are required to deliver sensed data

• Need to provide end-to-end delay guarantees• Sensors are limited in transmission capacity

and may suffer from low transmission reliability

Page 4: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Contributions of this Work

• Study the problem of providing end-to-end delay guarantee and throughput guarantee for multi-hop wireless sensor networks

• Develop scheduling policies for two kinds of networks

• Provide simulation results to justify the performance

Page 5: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Network Model

• A number of sensors transmitting data to a base station through multi-hop transmissions

• A routing tree is formed by the routing protocol, with the base station being the root

• h(n) = parent of n• h(6) = 4• h(5) = 2

98

76

4

5

3

2

1r

Page 6: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Traffic Model

• Time is slotted and grouped into intervals of length T time slots

• Each sensor may generate several flows• Packets generated in an interval need to be

delivered before the end of the interval, or they are dropped

T Flow 1 Flow 2 Deadline

Page 7: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Channel and QoS Model

• When a sensor n transmits to its parent, the transmission is successful with probability pn

• A flow f requires its throughput to be at least qf

• A scheduling policy is feasibility optimal if it satisfies requirements of all flows whenever feasible

Page 8: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Communication Model

• Consider two types of sensor networks

• Orthogonal relay system: Sensors can transmit and receive simultaneously– Sensors are equipped with full-duplex radio, or

they use OFDMA

• Half-duplex system: Sensors can either transmit or receive. They can receive one transmission at a time

Page 9: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Solution Overview

• Debt of flow f at interval k:

• Theorem: A policy that maximizes

in every interval is feasibility optimal

Indicator function of packet delivery

Page 10: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Orthogonal Relay System

• Greedy Forwarder: Each sensor transmits the packet with the largest debt among the available ones in each time slot

• Theorem: Greedy Forwarder is feasibility optimal for orthogonal relay system

Page 11: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Half Duplex System

• Closest Sensor First: Order packets by the number of hops between their current sensor and the base station, break ties by their debts

• Use this ordering to greedily select a maximal set of packets that can be transmitted simultaneously

• Theorem: Closest Sensor First is feasibility optimal for line topologies– Line topology: all flows are originated at the same

sensor

Page 12: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Simulation Setup

• 12 flows generated by sensors 3, 5, 6, 7, 8, 9• Channel reliability is randomly selected from

[0.4, 0.9]• Half of the flows require qf = α, others require qf = β

• Compare two policies– Random policy– Static priority

98

76

4

5

3

2

1r

Page 13: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Results for Orthogonal Relay Systems

00.10.20.30.40.50.60.70.80.9

1Greedy Forwarder

Static Priority

Random

β

α

Page 14: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Impact of Delayed Information

• Sensors notify their children information about debts periodically

• Sensors far away from the base station has stale information

00.10.20.30.40.50.60.70.80.9

1 Instant knowledgeUpdate period = 100Update period = 200

β

α

Page 15: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Results for Half Duplex Systems

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8Closest Sensor FirstStatic PriorityRandom

β

α

Page 16: Providing End-to-End Delay Guarantees for Multi-hop Wireless Sensor Networks

Conclusion

• Study the problem of providing end-to-end delay guarantees for wireless sensor networks with unreliable transmissions

• Develop scheduling policies for both orthogonal relay system and half duplex system

• They offer provable performance guarantees• Simulation results show that they are superior

than other policies