02 2 Networking Aspects

8/8/2019 02 2 Networking Aspects

http://slidepdf.com/reader/full/02-2-networking-aspects 1/38

[SelfOrg] 2-2.1

Self-Organization in AutonomousSensor/Actuator Networks

[SelfOrg]

Dr.-Ing. Falko Dressler

Computer Networks and Communication Systems

Department of Computer Sciences

University of Erlangen-Nürnberg

http://www7.informatik.uni-erlangen.de/~dressler/

[email protected]



[SelfOrg] 2-2.2

Overview

Self-Organization

Introduction; system management and control; principles andcharacteristics; natural self-organization; methods and techniques

Networking Aspects: Ad Hoc and Sensor Networks Ad hoc and sensor networks; self-organization in sensor networks;evaluation criteria; medium access control; ad hoc routing; data-centricnetworking; clustering

Coordination and Control: Sensor and Actor NetworksSensor and actor networks; coordination and synchronization; in-network operation and control; task and resource allocation

Bio-inspired Networking

Swarm intelligence; artificial immune system; cellular signalingpathways



[SelfOrg] 2-2.3

M AC Protocols for Ad Hoc and Sensor Networks

Principles and Classification

M AC A / M AC AW

S-M AC

Power Control M AC



[SelfOrg] 2-2.4

Principal Options and Difficulties

Medium access in wireless networks is difficult mainly because of

Impossible (or very difficult) to send and to receive at the same time

Interference situation at receiver is what counts for transmission success,

but can be very different to what sender can observe

High error rates (for signaling packets) compound the issues

Requirements

As usual: high throughput, low overhead, low error rates, «

Additionally: energy-efficient, handle switched off devices!



[SelfOrg] 2-2.5

Requirements for Energy-efficient M AC Protocols

Recall

Transmissions are costly

Receiving about as expensive as transmitting

Idling can be cheaper but is still expensive

Energy problems

Collisions ± wasted effort when two packets collide

Overhear ing ± waste effort in receiving a packet destined for another

node

Id l e list ening ± sitting idly and trying to receive when nobody is sending

P r otocol overhead

Always nice: Low complexity solution



[SelfOrg] 2-2.6

Design Issues

Distributed nature/lack of central coordination

Nodes must be scheduled in a distributed fashion

Exchange of control information

control packets must not consume too much of network bandwidth

Mobility of nodes

Very important factor affecting the performance (throughput) of the

protocol

Bandwidth reservations or control information exchanged may end up

being of no use if the node mobility is very high

Protocol design must take this mobility factor into consideration

system performance should not significantly affected due to nodemobility



[SelfOrg] 2-2.7

Classification of M AC Protocols

M AC Protocols for Ad

Hoc Wireless Networks

Contention-Based Protocols

Contention-Based

Protocols with Reservation

Mechanisms

Contention-Based

Protocols with

Scheduling Mechanisms

Other M AC Protocols

Sender-Initiated

Protocols

Receiver-Initiated

Protocols

Synchronous

Protocols

Asynchronous

Protocols

Single-Channel

Protocols

Multichannel

Protocols

M AC AW

F AM A

BTM A

DBTM A

RI-BTM A

M AC A-BI

HRM A

FPRP

M AC A/PR

RTM AC

DPS DLPS

MM AC MCSM A



[SelfOrg] 2-2.8


Cont ention-based pr otocols

No a priori resource reservation

Whenever a packet should be transmitted, the node contends with its

neighbors for access to the shared channel

Cannot provide QoS guarantees

Sender-i nitiat ed protocols±

packet transmissions are initiated by thesender node

n Single-channel sender-initiated protocols ± the total bandwidth is used

as it is, without being divided

n Multi-channel sender-initiated protocols ± available bandwidth is

divided into multiple channels; this enabled several nodes to

simultaneously transmit data

R ec eiv er-i nitiat ed protocols ± the receiver node initiates the contention

resolution protocol



[SelfOrg] 2-2.9


Cont ention-based pr otocols w it h reservation mec hanisms

Support for real-time traffic using QoS guarantees Using mechanisms for reserving bandwidth a priori

Synchronous protocols ± require time synchronization among all nodes inthe network global time synchronization is generally difficult to achieve

Asynchronous protocols ± do not require any global time synchronization,usually rely on relative time information for effecting reservations

Cont ention-based pr otocols w it h sc hedu ling mec hanisms

Focus on packet scheduling at nodes and also scheduling nodes for

access to the channel requirement for fair treatment and no starvation

Used to enforce priorities among flows

Sometimes battery characteristics, such as remaining battery power, areconsidered while scheduling nodes for access to the channel



[SelfOrg] 2-2.10

Contention-based Protocols: Main Problems

Hidden and exposed terminals - unique problem in wireless networks

H i dd en t erminal pr obl em±

collision of packets due to the simultaneous

transmission of those nodes that are not within the direct transmission

range of the sender but are within the transmission range of the receiver

Ex posed t erminal pr obl em ± inability of a node, which is blocked due to

transmission by a nearby transmitting node, to transmit to another node

S1 S2

R

R1 R2

S1 S2

Hidden terminal Exposed terminal



[SelfOrg] 2-2.11

Main Options to Shut Up Senders

Receiver informs potential interferers whil e a reception is on-going

By sending out a signal indicating just that

Problem: Cannot use same channel on which actual reception takes

place

Use separate channel for signaling

Bu sy tone protocol

Receiver informs potential interferers bef ore a reception is on-going

Can use same channel

Receiver itself needs to be informed, by sender, about impending

transmission

Potential interferers need to be aware of such information MAC A protocol



[SelfOrg] 2-2.12

BTM A ± Busy Tone Multiple Access

The transmission channel is split into

data and control channel

General behavior

When a node wants to transmit a packet,

it senses the channel to check whether

the busy tone is active

If not, it turns on the busy tone signal and

starts transmission

Problem: very poor bandwidth utilization



[SelfOrg] 2-2.13

M AC A ± Multiple Access Collision Avoidance

Use of additional signaling packets

Sender asks receiver whether it is able to receive a transmission - R eq u est to Send ( RT S) Receiver agrees, sends out a Cl ear to Send (C T S )

Sender sends, receiver acks

Potential interferers overhear RTS/CTS

RTS/CTS packets carry the expected duration of the data transmission

Store this information in a N et w ork Alloc ation Vector ( N AV)

Node 1

Sender

Receiver

Node 4

RTS

CTS ACK

D AT A

N AV

N AV

time



[SelfOrg] 2-2.14

M AC A ± Problems

RTS/CTS ameliorate, but do not solve hidden/exposed terminal

problems

Node 1

Node 2

Node 3

Node 4

RTS

CTS

D AT A

CTS

RTS

time



[SelfOrg] 2-2.15

M AC A ± continued

Collision handling

If a packet is lost (collision), the node uses the binary exponential back-off (BEB) algorithm toback off for a random time interval before retrying

Each time a collision is detected, the node doubles its maximum back-off window

Idle listening: need to sense carrier for RTS or CTS packets

In some form shared by many CSM A variants; but e.g. not by busy tones

Simple sleeping will break the protocol

MAC A pr otocol (used e.g. in I EEE 802.11)



[SelfOrg] 2-2.16

M AC AW Protocol

The binary back-off mechanism can lead to starvation of flows

Example

S1 and S2 are generating a high volume of traffic

If one node (S1) starts sending, the packets transmitted by S2 get collided

S2 backs off and increases its back-off window

the probability of node S2 acquiring the channel keeps decreasing

Solution

Each packet carries the current back-off window of the sender

A node receiving this packet copies this value into its back-off counter



[SelfOrg] 2-2.17

M AC AW Protocol

Large variations in the back-off values

the back-off window increases very rapidly and is reset after eachsuccessful transmission

Solution

multiplicative increase and linear decrease (MILD) back-off mechanism(increase by factor 1.5)

Fairness

M AC A: per node fairness

M AC AW: per flow fairness (one back-off value per flow)

Error detection

Originally moved to the transport layer Slow and introducing much overhead

Solution

New control packet type: data-sending (DS)



[SelfOrg] 2-2.18

M AC AW Protocol

Exposed terminal problem

RTS/CTS mechanism does not

solves the exposed terminal

problem

Solution

New control packet type: data-sending (DS), a small packet

(30 Byte) containing information

such as the duration of the

forthcoming data transmission



[SelfOrg] 2-2.19

Contention-Based Protocols with Reservation

M AC A/PR ± M AC A with Piggy-Backed Reservation

Multi-hop routing protocol based on M AC AW

Main components

M AC protocol

Reservation protocol

QoS routing protocol

Differentiation of real-time and best-effort packets General behavior

S lott ed mec hanisms

Maintenance of a reservation table (RT) at each node that records all thereserved transmit and receive slots / windows of all nodes within itstransmission range

Network allocation vectors (N AV) for cycles Destination sequenced distance vector (DSDV) used for routing

TDM-like system for real-time traffic

Best-effort traffic using M AC AW in free slots



[SelfOrg] 2-2.20

M AC A/PR Protocol



[SelfOrg] 2-2.21

M AC Protocol Using Directed Antennas

Properties

One receiver per node, which can transmit and receive only one packet atany given time

Each transceiver is equipped with M

directional antennas

Each antenna has a conical radiation

pattern spanning an angle of 2/M radians Basic RTS/CTS scheme (as used in M AC A)



[SelfOrg] 2-2.22

M AC Protocol Using Directed Antennas



[SelfOrg] 2-2.23

Power-Control M AC Protocol (PCM)

Properties

RTS/CTS are transmitted with maximum power pmax

RTS-CTS handshake to determine the required transmission power pdesired

RTS is received at the receiver with a signal level pr

Calculation of pdesired

R xthresh is the minimum necessary received signal strength

c « constant

c x p

p

p threshr

ax

desired *

!

measured

known in advance



[SelfOrg] 2-2.24

Power-Control M AC Protocol

RTS/CTSrange

1 2 3 6 7 8Data

transmission

D AT A/ ACK

range

4

carrier sensingrange

5

pmax pdesired



[SelfOrg] 2-2.25

Power-Control M AC Protocol

Properties

Adaptation to changing conditions, e.g. caused by mobility Instantaneous check and re-calculation of the necessary transmission power pdesired

Collision avoidance

Periodic bursts (after each EIFS) using pmax to notify neighbors about

ongoing transmissions



[SelfOrg] 2-2.26

Sensor-M AC (S-M AC)

Primary goal

To retain f l e x ibility of contention-based protocols while i mpr ov ing ener gy eff ici ency in multi-hop networks

(M AC A¶s idle listening is particularly unsuitable if average data rate is low - most of

the time, nothing happens)

Idea: Switch nodes off, ensure that neighboring nodes turn on simultaneously

to allow packet exchange (rendez-vous) Only in these acti ve per iod s, packet exchanges happen

Need to also exchange wakeup schedule between neighbors

When awake, essentially perform RTS/CTS

Coarse-grained sleep/wakeup cycle with duty cycle D = / T

time

Listen Sleep Listen Sleep

X

T



[SelfOrg] 2-2.27

S-M AC ± Scheduling

Use SYNC, RTS, CTS phases

Scheduling

Low-duty-cycle operation (1-10%)

All nodes choose their own listen/sleep schedules

These schedules are shared with their neighbors to make communication

possible between all nodes

Each node periodically broadcasts its schedule in a SYNC packet, which

provides simple time synchronization

To reduce overhead, S-M AC encourages neighboring nodes to adopt

identical schedules

time

Sync Data/Sleep

X

T

RTS/CTS Sync RTS/CTS



[SelfOrg] 2-2.28

S-M AC ± Synchronization

Nodes try to pick up schedule synchronization from neighboring nodes

If no neighbor found, nodes pick some schedule to start with

If additional nodes join, some node might learn about two different

schedules from different nodes

³Synchronized islands´

To bridge this gap, it has to follow both schemes

Complete algorithm

1. Listen for ³waiting time´ (at least one complete busy/sleep cycle) for

SYNC messages ± if nothing happens, the node chooses its own

schedule

2. If a node receives a SYNC bef ore setting up its own schedule, it takesover the received schedule

3. If a node receives a SYNC af t er setting up its own schedule, its adopts

both schedules to bridge two islands



[SelfOrg] 2-2.29

S-M AC ± Synchronization

S1 S1Start: Node 1

Waiting time

R1 S1Start: Node 2

S4 S4Start: Node 4

Waiting time

R1 S4Start: Node 3

Abbreviatedwaiting time

R4

Abbreviated

waiting time

Adapted sync

Adapted sync

Adapted sync

S1

S1

S1

time



[SelfOrg] 2-2.30

S-M AC ± Performance Aspects

Standard S-M AC

Energy saving through periodic sleep

Depending on the duty cycle, the end-to-end performance is increasing as

n Per busy period, exactly one packet can be transmitted within a

common radio range

n If rather short packets need to be transmitted either long sleep

intervals must be prevented (energy wastage) or the per-hop delay isfurther increased

Improved S-M AC

Ad apti ve list ening allows additional energy savings (nodes wake up

immediately after the exchange completes for immediate contention for

the channel)



[SelfOrg] 2-2.31


Standard S-M AC w/o adaptive listening

S R/C

Data

Sleep

S R/C

Data

S R/C

Data

Sleep

C

TimeListen/Sleep

R

C A

Sleep

Sleep

Sleep

Slot n Slot n+1 Slot n+2

S Sync R/C RTS/CTS R RTS C CTS A ACK

Listen/Sleep

R

C A

Sleep

Sleep

Sleep

Sleep

Listen/Sleep

R

C A

Sleep

Sleep

Sleep

A

B

D



[SelfOrg] 2-2.32


Improved S-M AC w/ adaptive listening

A

B

C

S R/CTime

R

C

Data

A Sleep

Slot n Slot n+1 Slot n+2

S Sync R/C RTS/CTS R RTS C CTS A ACK

S R/C

R

C

Data

A Sleep

Sleep

S R/C

R

C

Data

A

Sleep

Sleep

Sleep

Sleep

ALP ALP

Adaptive Listening ALP

D

Sleep

Sleep

Sleep

Sleep

Sleep



[SelfOrg] 2-2.33

S-M AC ± Performance Evaluation

Experimental setup

Ten nodes in a line

Analyzed S-M AC modes

Mode1: no periodic sleep (= M AC A)

Mode2: 10% duty cycle, w/o adaptive listening (= standard S-M AC)

Mode3: 10% duty cycle, w/ adaptive listening (= improved S-M A

C)

1 2 3 8 9 10«

source sink



[SelfOrg] 2-2.34


M ean energy consumption per byt e ± the total energy consumed by all

nodes divided by the total number of bytes received by the sink



[SelfOrg] 2-2.35


E nd -to-end good put ± the total number of bytes received by the sink

divided by the time from the first packet generated at the source untilthe last packet was received by the sink



[SelfOrg] 2-2.36


M ean end -to-end delay ± the sum of all end-to-end delays divided by

the total number of packets



[SelfOrg] 2-2.37

Summary (what do I need to know)

Well-established M AC protocols in the ad hoc domain

M AC A / M AC AW / 802.11 Similar solutions for hidden/exposed terminal problem

Applicability for wireless sensor networks

S c al ability ± M AC A/802.11 needs a global sync; adaptive solutions aredemanded

E ner gy eff ici ency - limited sleeping time in M AC A/802.11; low dutycycles and/or adjustments of the transmission power are needed

Specific developments

P C M ± well-controlled transmission power, can be combined with any

RTS/CTS based M A

C protocol S - MAC ± supports multiple schedules and long sleep cycles with adaptive

listening



[SelfOrg] 2-2.38

References

V. Bharghavan, A. Demers, S. Shenker, and L. Zhang, "M AC AW: A Media Access

Protocol for W

ireless L A

N's," Proceedings of A

CM SIGCOMM'94, London, UK,September 1994, pp. 212-225.

P. Karn, "M AC A: a new channel access method for packet radio," Proceedings of

ARRL/CRRL Amateur Radio 9th Computer Networking Conference, London, Ontario,

Canada, 1990, pp. 134-140.

IEEE, "Wireless L AN Medium Access Control (M AC) and Physical Layer (PHY)Specification," IEEE Std. 802.11-1999 edition, 1999.

E.-S. Jung and N. Vaidya, " A Power Control M AC Protocol for Ad Hoc Networks,"Proceedings of ACM/IEEE MobiCom, September 2002.

W. Ye, J. Heidemann, and D. Estrin, " An Energy-Efficient M AC Protocol for WirelessSensor Networks," Proceedings of 21st International Annual Joint Conference of theIEEE Computer and Communications Societies (INFOCOM), vol. 3, New York, NY,US A, June 2002, pp. 1567-1576.

W. Ye, J. Heidemann, and D. Estrin, "Medium Access Control with Coordinated Adaptive Sleeping for Wireless Sensor Networks," IEEE/ACM Transactions on

N etworki

ng (TON), vol. 12 (3), pp. 493-506, June 2004.

F. Chen, F. Dressler, and A. Heindl, "End-to-End Performance Characteristics inEnergy- Aware Wireless Sensor Networks," Proceedings of Third ACM InternationalWorkshop on Performance Evaluation of Wireless Ad Hoc, Sensor, and UbiquitousNetworks ( ACM PE-WASUN'06), Torremolinos, Malaga, Spain, October 2006, pp. 41-47.

Documents

02 2 Networking Aspects