Transcript Design Challenges For Energy-Constrained Ad-Hoc

Design Challenges For EnergyConstrained Ad Hoc Wireless Networks
Andrea J. Goldsmith, Stephen B. Wicker
IEEE Wireless Communications, August 2002.
2006. 11. 20
Summarized by Lee Chulki, IDS Lab., Seoul National University
Presented by Lee Chulki, IDS Lab., Seoul National University
Contents
 Introduction
 Applications
 Cross-Layer Design
 Conclusions
2
Introduction
 Ad hoc wireless network

Peer-to-peer communication

Distributed networking and control functions among all nodes

Multihop routing
 Don’t think that it must be completely flat

The distinguishing emphasis in the ad hoc approach lies in the
design requirements
 Energy constraints are not inherent to all ad hoc wireless
network

But some of the most exciting applications are in energyconstrained category
3
Applications


Data Networks

Data exchange between laptops, PDA…

Ex) LAN
Home Networks



Device Networks

Replace inconvenient cabled connections with wireless connections

Ex) Bluetooth
Sensor Networks


Using PDA in the bedroom to scan music in PC
With Non-rechargeable battery / minimize human intervention
Distributed Control Systems

Remote plants, sensors and actuators linked together via wireless
communication channels

Ex) Automated Highway System
4
Cross-Layer Design
 Layered approach

Simplified network design

Led to the robust scalable protocols in the Internet
 Problem

Inflexibility and suboptimality
–

A wide range of network requirements / energy constraints
Poor performance for ad hoc wireless networks
–
With Energy constraint, high bandwidth needs, delay constraints…
 Need Cross-Layer Design

Supports adaptivity and optimization across multiple layers
5
Cross-Layer Design

Link Layer


MAC Layer


Adapt based on link and interference conditions, delay constraints, bit
priorities
Network Layer


Adapt rate, power and coding to meet the requirements of the application
Use adaptive routing protocols based on current link, network and traffic
conditions
Application Layer

Utilize a notion of soft QoS

Adapts to the network conditions to deliver the highest possible application
quality
6
Cross-Layer Design
7
Cross-Layer Design
 Two fundamental questions

What information should be exchanged across protocol layers and
how should that information be adapted to?

How should global system constraints and characteristics be
factored into the protocol designs at each layer?
 Discuss the design of the different layers

Link

MAC

Network

Application
8
Link Design Issues
 Goal

Achieve rates close to the fundamental capacity limits of the
channel while overcoming channel impairments using relatively little
energy
 Contents

Fundamental Capacity Limits

Coding

Multiple Antennas

Power Control

Adaptive Resource Allocation
9
Fundamental Capacity Limits

The maximum data rate that can be transmitted over the channel with
arbitrarily small probability of error

Researches

The capacity of an AWGN (Additive White Gaussian Noise) Channel
–



with B (Bandwidth), SNR (Signal-to-Noise power Ratio)
Recent works: for models that better reflect underlying current wireless
system
More concepts

Capacity per unit energy

Capacity in bits
With finite energy -> Can transmit finite bit

Information transmission

Exchange of routing information

Forwarding bits for other nodes
10
Coding

Reduce the power required to achieve a given Bit Error Rate

Researches

Family of codes on graphs with iterative decoding algorithms
–
Ex) Turbo code
–
Require more signal processing power
11
Multiple Antennas
 Improve the performance
 Reduce transmit power
 Categories

Diversity

Beamsteering

MIMO (Multiple Input Multiple Output)
 Trade-off

Save transmission power

Often require significant power for signal processing (complexity)
12
Power Control
 Potent mechanism for improving wireless network performance
 Strategies

Maintain SINR on the link above a required threshold by increasing
power relative to fading and interference
–
Works well for continuous stream traffic with a delay constraint
–
Not power-efficient

Dynamic programming to minimize the transmit power required to
meet a hard delay constraint

…
 Significant impact on protocols above the link layer

The level of transmitter power defines the “local neighborhood”
13
Adaptive Resource Allocation
 Provides robust link performance with high throughput while
meeting application-specific constraints.

A relatively new technique
 Researches

Combinations of power, rate, code, and BER adaptation

Variation of the link layer retransmission strategy as well as its
frame size

Diversity combining of retransmitted packets or retransmitting
additional redundant code bits instead of the entire packet
14
Medium Access Control Design Issues
 Goal: How different users share the available spectrum?

Divide the spectrum into different channels

Assign these different channels to different users
 Contents

Channelization

Random Access

Scheduling

Power Control
15
Channelization
 Frequency division

The system bandwidth is divided into non-overlapping channels

Simple but inflexible
 Time division

Time is divided into orthogonal time slots

More flexible than frequency division
 Code division

Time and bandwidth are used simultaneously by different users,
modulated by orthogonal or semi-orthogonal spreading codes
 Hybrid combinations

Combinations of above methods
16
Random Access

Assign channels to active users (not to idle users)



Most systems have many more total users than channels
Collision can be reduced by CSMA (Carrier Sense Multiple Access)

Make hidden / exposed terminal problem

Solutions: 4-way handshake / busy tone transmission / hybrid techniques
Researches

Sleep: more energy-efficient

Dynamic programming approach to decisions about transmissions
–
More flexible and more energy aware
17
Scheduling
 Random access protocols
 Good to bursty traffic
 Poor to long strings of packet or continuous stream data
 Solution: not easy
 Distributed scheduled access in ad hoc wireless networks in
general is an NP-hard problem.
 Researches
 PRMA (Packet Reservation Multiple Access)
–
Combines the benefits of random access with scheduling
 Optimal scheduling algorithms to minimize transmit energy
18
Power Control
 Researches

Maintaining the SINR of each user sharing the channel above a
given threshold
–
Performed in a distributed manner

Strategy for multiple access that takes into account delay
constraints

For cellular systems
–
Centralized / distributed power control
 This issue remains Active area of research
19
Network Design Issues
 Contents

Neighbor Discovery And Network Connectivity

Routing

Scalability and Distributed Protocols

Network Capacity
20
Neighbor Discovery And Network Connectivity
 Neighbor discovery

Higher transmit power, more neighbors

Require larger neighborhoods for high mobility
 Connectivity

Influenced by the ability to adapt parameters at the link layer
–
Such as rate, power, coding
 Sleep decisions are important

Network connectivity

Neighbor discovery
21
Routing
 Multihop routing protocols

Flooding, centralized/distributed proactive routing, reactive routing

Combination of reactive and proactive routing
 Mobility

Flooding is effective under high mobility

Multipath routing: modification of flooding
–
A packet is duplicated on only a few paths with a high likelihood of
reaching its final destination
 Energy constraints

Reactive routing is effective

With listening mode, proactive and reactive routing have roughly
the same energy consumption
22
Scalability and Distributed Protocols



Scalability

Important in the design of self-configuring ad hoc wireless networks

Most work on scalability has focused on small networks (<100 nodes)
Distributed network control algorithms

The key to self-configuration

Often consume a fair amount of energy in processing and exchange
Researches

Self-organization

Distributed routing

Mobility management

QoS support

Security
23
Network Capacity
 Fundamental capacity limit of an ad hoc wireless network
 Researches

The per-node rate in a large ad hoc wireless network goes to zero
–
Even with optimal routing and scheduling
–
So, All nodes should not communicate with all other nodes

Node mobility actually increases the per-node rate to a constant

Determine achievable rate regions using adaptive transmission
strategies

Information theoretic analysis on achievable rates between nodes
24
Application Design Issues
 Adaptive QoS

Unrealistic: low-capacity, mobile users, dynamic topology…

Applications must adapt to time-varying QoS parameters offered by
the network
–
Ex) Rate-Delay trade-off curve: Decide point to operate
 Application Adaptation

Ex) Video: Change compression rate

Demanding applications can deliver good overall performance under
poor network conditions if the application is given the flexibility to
adapt
25
Conclusions
 Cross-layer design is particularly important under energy
constraints

Energy across the entire protocol stack must be minimized
 Out-of-box thinking is required

The box of layered protocol designs

The box of wireline protocols

The box of guaranteed QoS for demanding applications
26