• MAC Essentials for Wireless Sensor

Networks
• Abdelmalik Bachir, Mischa Dohler, Senior Member, IEEE, Thomas
Watteyne, Member, IEEE, and
Kin K. Leung, Fellow, IEEE
 Medium access control

• Part of the link layer (OSI model)

• Controls access to physical medium

• Assigns a unique address to the host (MAC address)

Why is MAC important in WSNs?
 Some WSN characteristics
• Limited energy resources

• Low data load , but highly directed

• Control often decentralized

• Huge amount of nodes with low computing capabilities

• Volatile radio links

 Energy constraints in WSN

• Energy consumption
dominated by nodes radio
consumption

• MAC controls active and

sleep mode

• MAC essential in energy

consumption of a WSN
design
 Energy consumption
• Idle listening

• Collisions

• Overhearing

• Large overheads
 Traditional MAC families

• Reservation-based
protocols

• Contention-based
protocols
 Reservation-based protocols

• Nodes access network in a

scheduled order

• Global time synchronization

• High fixed troughput

• Poor mobility

• TDMA
 Contention-based protocols
• Nodes access the network in a random and competitive
order (e.g channel sensing and contention)

• Robust to node mobility

• Degraded troughput with heavy loads

• CSMA , ALOHA
Throughput vs offered load
Energy consumption vs throughput
 Thematic taxonomy
• Traditional surveys classify
MAC protocols according to
reservation or contention
based medium access
techniques

• Need guidelines to determine

MAC protocols suited for a
given set of cirumstances

• Focus on thematic
classification of protocols
suited for specific needs of
different WSN
 Proposed families
• Scheduled protocols:
Fixed high load traffic

• Protocols with Common active periods:

Periodic medium load traffic

• Preamble sampling protocols:

Random low load traffic
 Scheduled protocols
• Optimized for high-load WSNs
e.g multimedia applications

• Maintains network wide

schedule among nodes

• No collisions , no overhearing
and minimized idle listening
reduce energy consumption

• Weak node mobility

• Requires global
synchronization
 Basic functionality

• Nodes assigned slots inside

timeframes

• Slots assigned using a

distributed schedule

• Network-wide synchronisation
and scheduling needed

• TDMA
 Time synchronized Mesh Protocol

• A node can participate in multiple frames at once having

multiple refresh rates for different tasks

• TDMA combined with FDMA and frequency hopping

• Increased robustness against narrowband interference

Time Synchronised Mesh protocol
 Some related problems
• Maintining tight synchronization in multi-hop networks

• Repetitive broadcast may waste energy

• Poor flexibility: Changes in traffic load or network

structure require new schedules to be calculated

• Maintaining memory of neighbourhood topology

consumes energy inside nodes
 Protocols with Common Active
Periods
• Suited for periodic medium-load
traffic (typical to e.g industrial
applications)

• Common active periods among

nodes to reduce energy
consumption

• Local self-scheduling among

nodes increase network flexibility

• Weak against node mobility

• Some degree of synchronization

needed
 Basic functionality
• Nodes define common active/sleep periods

• Nodes contend for channels during active periods

• Synchronization among nodes to define active periods

 Sensor MAC (SMAC)
• Active period split into two sub periods (Sync and data)

• Sync packets exchanged to build local schedule

• Nodes with common schedule form virtual clusters

• Sub periods divided into mini-slots where nodes contend

for channel

• Random backoff time to reduce collisions

 Some related problems
• Determining optimal size of active periods (Short periods
reduce idle listening but increase collisions)

• Long sleep periods increase network latency

• Irregular traffic increase collisions during active periods

• Mobile nodes waste energy rescheduling

 Preamble sampling protocols
• Good for WSN’s with low traffic
and remote node locations (e.g
metering applications,
environmental monitoring)

• Long sleep cycles reduce

energy consumption except
during transmission

• Robust against node mobility

• Requires few synchronization

among nodes
• Sleeping node switch on radio for short durations to
sample channel (check intervals)

• Transmitter uses long preamble to ensure detection by

recieving node during check intervals
 Some related problems

• Transmitter preamble needs to cover entire reciever

check time interval to ensure detection

• Long preamble drains transmitter energy and increase

chance of collisions when network traffic increases

• High cost of transmission can drain more energy than

saved from reduced listening
Finding optimal check intervals depend on
traffic load
Reducing preamble length without reducing
check interval: Wireless sensor MAC
(WiseMAC)
• Node learns about neighbours checktimes to reduce
preamble

• Recieving nodes share check-times using piggybacking

on ACK returned to transmitter

• Neighbours check times stored in nodes internal table

• If reciever check times change, transmitter can switch

back to full preamble length to ensure detection
Wireless sensor MAC (WiseMAC)
 Hybrid protocols
• Combine categories to achieve high
performance in variable traffic patterns

• Zebra MAC: CSMA inside large TDMA

slots
 Conclusions
• MAC important in WSNs

• Complete system-wide quantificaiton of WSN constraints

not available

• No MAC exist that is highly scalable

• Power harvesting in future applications may favour some

scheduling of nodes active modes
 Debate
• Will real world systems with varying traffic, multi
hops etc, result in too many hybrids making the
families useless?

• Making spesific families linked to real world

systems may encourage thinking inside a box
and hinder seeing new solutions

• Difficult to see practical difference between

reservation based and scheduling protocol
family
T-Lohi: A New Class of MAC Protocols for
Underwater Acoustic Sensor Networks

Affan A. Syed Wei Ye John Heidemann

Information Sciences Institute/USC
 Paper motivation
• No MAC protocol widely available to
support dense UWSNs

• Unique characteristics of underwater

communication may be exploited in MAC
design

• Energy-efficient design important in

UWSNs
 Paper main contributions

• Evaluation of space time uncertainty and

latency and how these may be exploited in
MAC design

• T-Lohi, a new class of MACs for UWSNs

 Underwater Acoustic sensor
networks
• Traditional networks very long range
(several kilometers)

• Many new limitations not seen in radio

• Short range acoustic networks may be

easier to implement
 Challenges with underwater
acoustic communication
• High propagation latency

• Low bandwidth

• Surcface scattering

• Complex multipath fading

• Temperature, salinity and

pressure influence
 Challenges with underwater
acoustic communication
• Short range communication (less than 500m) may
simplify propagation characteristics and allow simpler
and cheaper design

• Transmission often 100 times more expensive than

reception

• These characteristics create new-phenomena in MAC

protocol design
 Energy reliable applications
• Static sensing applications

• Gliders and low-energy mobility platforms

• Water-life tagging
 Energy efficiency design
challenges
• High transmit costs

• Long idle listening time

• Replace battery may be difficult

 Space time uncertainty
• High latency cause unfairness in contention due to
node locations

• Clear channel assesment needs to consider worst case

latency of neighbours

• Nodes in close proximity may monopolize a channel

• High latency may be exploited in counting and detecting

contenders
 Deafness conditions
• Nodes cannot transmit and recieve simultanously (half-
duplex)

• Two contending nodes far away may not hear

eachothers contending tone and think they won round

• Nodes may start contending while recieving other

contenders and not hear

• Short reciever detection time important to reduce

deafness probability
 Deafness conditions impact
• Tone-data collisions

• Data-data collisions

• Increasing number of
conteders reduce risk
of deafness
T-Lohi MAC protocol
Tone-LOHI MAC protocol goals
• Efficient channel utilization

• Stable troughput with both low and high loads

• Low energy consumption

• Flexible for a range of applications and traffic

patterns
 Tone-LOHI MAC protocol
• Nodes reserve channel prior to data
transmission using contention rounds

• Contention rounds may be common or random

among nodes
 T-LOHI energy reduction
• Channel reservation prevents packet collisions

• Special wake-up tone reciever allowing low-

power wakeup listening
 T-LOHI channel reservation
• Nodes contend to reserve channel

• Nodes send short tone and wait to listen if

channel clear

• If not clear, count number of contenders

and decide backoff time
T-Lohi Frame structure
 T-LOHI Data transfer
• Transmitter responsible for waking up
recievers

• Reciever of wake-up tone scan data

channel for preamble

• If preamble, scan data header to check if

correct recipient
 T-LOHI tone implementation
• Custom, low power tone reciever

• Share channel with data

• Consume 1/100th the energy of listening

for data

• Can be replaced by short data packets

 False tone detection
• Channel noise may sound like tone

• Noise tone may be interpreted as false

contender and prolong contention rounds

• Longer contention rounds reduce troughput

• Low energy cost of contention give false tone

detection small impact on energy consumption
 T-LOHI Flavours
• Synchronized T-Lohi (ST-Lohi)

• Conservative Unsynchronized T-Lohi

(cUT-Lohi)

• Aggressive UT-Lohi (aUT-Lohi)

 Synchronized T-Lohi (ST-Lohi)

• Contention rounds syncronised

among nodes

• Observe channel for worst case

propagation time

• Count contenders and estimate

distances

• No bidirectional deafness

• Nodes close to reciever higher

chance of backoff(SAI)
 Conservative Unsynchronized T-
Lohi (cUT-Lohi)
• Random contending from nodes

• Observe channel for twice the worst case

propagation time to avoid collision

• Simpler to implement

• Cannot count number of contending nodes

 Aggressive UT-Lohi (aUT-Lohi)

• Random contending from nodes

• Half the contention rounds of cUT-Lohi

• Shorter rounds give higher througput

 Protocol weakness
• Tone-data collisions may occur in aUT-Lohi
due to random low contention periods

• Data-data collisions may occur from

bidirectional deafness

• Adding more contenders reduce risk of

bidirectional deafness
 Performance evaluation
• Performance of protocols tested in a network simulator

• Packet loss due to channel noise and multi path not

accounted for

• Results show througput increase of 34-50% of

comparable underwater MAC protocols
 Channel utilization with increasing
load
• Protocols show high efficiency at low load

• Stable throughput at about 50% of channel capacity at

heavy loads

• ST-Lohi and aUT-Lohi offer higher throughput due to

smaller contention rounds

• ST-Lohi throughput not affected by increasing network

density
 Energy efficiency
• Energy eff. Measured as cost beyond optimal energy per
packet during transmission (Overhead)

• ST-Lohi show very low and constant overhead (4%) due

to collision prevention mechanisms

• aUT-Lohi show highest overhead at heavy loads due to

increased collisions

• cUT-Lohi has longer sleeping periods during

transmission so energy cost per packet becomes similar
to aUT-Lohi
Energy efficiency
 Packet loss
• Increase in network density reduce collisions in aUT-Lohi

• More contenders reduce risk of deafness conditions and

data-data collisions

• cUT-Lohi provide most robust and reliable data transfer

for sparse and low traffic networks
Packet loss
Impact of counting conteders on
fairness
 Conclusions

• T-Lohi offer good performance in fully connected

network. Need studies of multi-hop networks

• All flavours within 3-9% of optimal energy efficiency

• All flavours within 30% of optimal channel utilization

Discussion