Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Increased QoS through a Degraded Channel using a Cross-Layered HARQ Protocol

Elliot Ranger

Brad Gaynor

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Topics

• Review• Project Overview• Simulation Plan• Simulation Execution• Future Work

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Previous Comments

• Won’t the protocol use more power to form a cluster?– Yes, but link integrity will increase– May result in lower energy consumption to the network

• Less energy drain on a single node

• More efficient routes discovered faster

– But, will add overhead to the overall network (congestion)

• Where is the cluster intelligence?– Distributed in each node

• Nodes query their neighbors for assistance

• How/When does the cluster dissolve?– When a better route is discovered

• One node always gets the correction through

• Other metrics

• When do you stop sending FECs?– Same as existing HARQ protocols

A B C D

L = LowSNR = Low

E

F

Noise

L = HighSNR = Low

L = HighSNR = High

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Review

• Take advantage of localized temporal uncertainty– A lower probability link may have

a better chance of success at any given time

• A lower probability link may be incorrectly labeled as such

• Varying network topology • Localized noise characteristics

• Send different FEC from multiple sources – Increases probability of receiving the correct message

– Multiple transmit nodes relieve the strain on a single node

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Thesis

Thesis: HARQ using multiple, collaborative nodes results in increased probability of message reception and extends the overall lifetime of the network.

A B C D

Noise

E

F

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Project Overview

• Implementation– Prior work

• HARQ• Turbo Codes

– Cross-layer support• Network Protocol• Network/Data-link layer interface

• Simulation– Model network

• Topology• Noise• Mobility

– Simulate implementation

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Simulation Scenarios

• Sim 1 - Simple Topology– Fixed Transmitter– Mobile Receiver– Results: errors due to path loss

• Sim 2 - ARQ– Same Network Topology– Add Automatic Repeat Request

• Sim 3 – Multiple Nodes– Add our approach of multiple

collaborative nodes

• Sim 4 – HARQ– Add HARQ protocol

• Sim 5 – Complete Simulation– Our approach on a complete

mobile ad-hoc network – Many nodes and events

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Sim 1 – Node & Process Models

• Node Models– Transmitter

• Packet Generator

• Radio Transmitter

• Antenna

– Receiver• Antenna

• Radio Receiver

• Packet Sink

• Process Model– Simple Source

• Generates Packets

• Interval Time

• Size of Packet

• Sends to Transmitter

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Sim 1 – Network Model

• Network Model– Describes Network Topology– Fixed Transmitter– Mobile Receiver

• Trajectory (Green Arrow)

– Noise Model

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Sim 1 – Simulation Results

• Signal-to-Noise Ratio– Decreases as mobile receiver

moves away from transmitter– Path Loss is modeled as 1/R2

• Bit Error Rate– Increases as mobile receiver

moves away from transmitter

Tufts University. EE194-WIR Wireless Sensor Networks. March 3, 2005

Future Work

• In progress– Building a simple ACK/NACK protocol– Modeling noise & channel characteristics– Simulation 2+

• Statistical models– Noise (Gaussian)– Bit Error (Model for receiver sensitivity)– Packet Loss (S/N + FEC gain)

• Mathematical models– Write models to describe the theory we are attempting to prove in simulation