Embed Size (px)
Citation preview
Network Embedded Systems
Sensor Networks
Physical and Link Layer
Marcus Chang, [email protected]
Slides: Andreas Terzis
1
Physical Layer
2 Image: Dino.korah
PHY Layer Motivation
Nodes communicate over wireless channel
Characteristics of wireless channel and transceiver have impact on all upper layer protocols
Transmission range
Loss rate
Energy consumption
Why?
Understand performance of existing systems
Guide design of future systems
3
Communication System
Fundamental parameters that control rate and quality of information exchange are the channel bandwidth B and the signal power S
Bandwidth: range of frequencies that the channel can transmit
Increasing signal power reduces the effect of channel noise
4
Signal Types Analog signals
Discrete signals can be generated by sampling analog signals
Digital signals
Information signal (baseband) Example: sequence of bits we want
to transmit Low frequency signals do not travel
well through most mediums such as wires, and wirelessly through air.
To transmit them over long distances requires a form of modification called modulation
Carrier signal: A carrier is a pure sinusoid of one
particular frequency and of a certain phase.
5
Example: Amplitude Modulation (AM)
Carrier: sin(2πfct)
Signal: I(t)
Signal modulates the carrier’s amplitude
I(t)×sin(2πfct)
6
Modulating Digital Signals
Similar concept but information signal is digital
Amplitude Shift Keying (ASK)
Frequency Shift Keying (FSK)
Phase Shift Keying (PSK)
Can use multiple amplitudes, frequencies, phases
Combinations of amplitude and phase manipulations (e.g., QAM)
7
Example: 802.15.4
Modulation
4-bit symbols are mapped to one of 16 pseudo-random 32-chip sequences
Modulation in the 2.4 GHz range is Offset-Quadrature Phase Shift Keying (O-QPSK) with half-sine chip shaping
Demodulation
Incoming 32-chip sequences are mapped to the symbol with the shortest Hamming distance
Redundancy enables error correction
8 Image: Texas Instruments CC2420 Datasheet
Bit Error Rate
Noise can corrupt received signal to the point that it is incorrectly decoded
Bit errors
For a given modulation scheme, bit error rate (BER) is a function of Signal-to-Noise ratio (SNR)
9
10
Signal to Noise Ratio
Noise of power N0 arrives at the sensor
SNR= Pr / N0
Usually measured in decibels: SNRdB= 10logSNR
When SNR drops below threshold signal cannot be detected
How can we increase detection range?
Quadrupling the power doubles the range
Double the carrier wavelength
Improve detection algorithm
11
Propagation Medium Losses
Propagation medium usually introduces distortion, scattering, and attenuation
Propagation loss LP
Absorption in uniform medium is a constant fractional loss per unit distance
Example: Fading due to rain as function of frequency
F(GHz) Lp (dB/km)
7.5 0.048
9.4 0.093
16 0.4
34.9 2.34
12
Obstructions
Obstructions, media changes, and reflective objects introduce further losses
Reflection
Refraction
Diffraction
Scattering
13
Log-Normal Shadowing Model
Assume average power (in dB) decreases proportional to the log of distance
Path-loss exponent n, depends on propagation environment
Environment n
Free Space 2
Urban Area 2.7-3.5
Shadowed Urban 3-5
In-building LOS 1.6 to 1.8
Obstruction in building 4 to 6
Obstruction in factories 2 to 3
PL(d) PL(do)10n logd
d0
14
Log-Normal Shadowing Model
dB)(in deviation
standard with dB)(in r.vGaussian mean -zero is
X
PL(d) PL(do)10n logd
d0
Statistically describes random shadowing effects
values of n and σ are computed from measured data using linear regression
Model typically derived from measurements
Log normal model found to be valid in indoor environments
Describe signal strength variations over large distances and large time scales
Telosb Radio – Texas Instruments CC2420
15
Link Layer
16
Breaking the Stack
2002 2003 2004 2005 2006 2007 1999 2000 2001 2008
Link
Network
Transport
Application S
PIN
Bro
adcast S
torm
Tin
yO
S 0
.6
Wo
oM
AC
GA
F
S-M
AC
Alo
ha P
S
PS
FQ
GD
I
CO
DA
Min
tRo
ute
T-M
AC
RM
ST
B-M
AC
Mu
ltih
opLQ
I
Fu
sio
n
De
luge
FP
S
Wis
eM
AC
SP
Dri
p
Dra
in
BV
R
Re
dw
oods
No
rth
Se
a
SC
P-M
AC
P
ED
AM
AC
S
X-M
AC
Vo
lcano
CT
P
Flu
sh
Go
lden G
ate
Brid
ge
S4
Dip
17
MAC and its Classification
Medium Access Control (MAC)
When and how nodes access the shared channel
Classification of multiple access MAC protocols
Scheduled protocols
Schedule nodes onto different sub-divisions
Examples: Time (TDMA), Frequency (FDMA), Code (CDMA)
Contention-based protocols
Nodes compete in probabilistic coordination
Examples: ALOHA (pure & slotted), Carrier Sense (CSMA)
18
19
MAC Attributes
Collision avoidance
Basic task of a MAC protocol
Energy efficiency
Scalability and adaptivity
Network size, node density and topology change
Channel utilization
Latency
Throughput
Fairness
Primary
Secondary
20
Energy Efficiency in MAC Design
What causes energy waste?
Packet collisions
Control packet overhead
Overhearing unnecessary traffic
Long idle time
Bursty traffic in sensornet applications
Idle listening consumes 50—100% of the power for receiving
Terminology
Wakeup period (time between wakeups)
Duty Cycle = listen period/Wakeup period
21
Scheduled Protocols
TDMA
Advantages
No collisions
Energy efficient — easily support low duty cycles
Disadvantages
Bad scalability and adaptivity
Difficult to accommodate node changes
Difficult to handle inter-cluster communication
Requires time synchronization
22
Polling
A special TDMA without pre-assigned slots
A master plus one or more slaves (star topology)
The master node decides which slave can send by polling the corresponding slave
Only direct communication between the master and a slave
Examples
IEEE 802.11 infrastructure mode (CFP)
Bluetooth piconets
Scheduled Protocols
23
CSMA — Carrier Sense Multiple Access
Listening before transmitting
Collisions can still occur
Examples
IEEE 802.11 – CSMA/CA
Collision Avoidance – random back-off time
Problem
Solution Explicit Request-to-Send and Clear-to-Send (RTS/CTS) packets
Contention-Based Protocols
a b c
Hidden terminal: a is hidden from c’s carrier sense
24
Case Study: S-MAC
S-MAC — by Ye, Heidemann and Estrin
Tradeoffs
Increase latency and decrease fairness to improve energy efficiency
Major components in S-MAC
Periodic listen and sleep
Collision avoidance
Overhearing avoidance
Message passing
From “Medium Access Control With Coordinated Adaptive
Sleeping for Wireless Sensor Networks” by Ye et al.
25
Coordinated Sleeping
Problem:
Idle listening consumes significant energy
Solution:
Periodic listen and sleep
Turn off radio when sleeping
Reduce duty cycle to ~ 10% (120ms on/1.2s off)
sleep listen listen sleep
26
Coordinated Sleeping
Schedules can differ
Nodes prefer neighboring nodes that have same schedule
Border nodes: two schedules
Node 1
Node 2
sleep listen listen sleep
sleep listen listen sleep
Schedule 2
Schedule 1
27
Coordinated Sleeping
Schedule Synchronization
New node tries to follow an existing schedule
Remember neighbors’ schedules
— to know when to send to them
Each node broadcasts its schedule every few periods of sleeping and listening
Re-sync when receiving a schedule update
Periodic neighbor discovery
Keep awake in a full sync interval over long periods
28
Collision Avoidance
S-MAC is based on contention
Similar to IEEE 802.11 ad hoc mode (DCF)
Physical and virtual carrier sense
Randomized backoff time
RTS/CTS for hidden terminal problem
RTS/CTS/DATA/ACK sequence
29
Adaptive Listening
Reduce multi-hop latency due to periodic sleep
Wake up for a short period of time at end of each transmission
Reduces latency by half
4 1 2 3
CTS
RTS
CTS
listen listen
t1 t2
receive send Node 3:
t0
30
Overhearing Avoidance
Problem: Receive packets destined to others
Solution: Sleep when neighbors talk
But only use in-channel signaling (RTS/CTS)
Who should sleep?
All immediate neighbors of sender and receiver
How long to sleep?
The duration field in each packet informs other nodes the sleep interval
31
Adaptive Listen Slots
In S-MAC all nodes have listen slots of the same duration
Different nodes might have different Tx/Rx patterns
Idle listening wastes power
Idea: adaptively change the idle listen slot
“An Adaptive Energy-Efficient MAC Protocol for Wireless Sensor Networks” (aka T-MAC) by T. van Dam, K. Langendoen
Principle
Node periodically wakes up, turns radio on and checks channel
Wakeup time fixed, “Check time” variable
If energy is detected, node powers up in order to receive the packet
Noise floor estimation used to detect channel activity during LPL
Low Power Listening (B-MAC)
32 From “Versatile Low Power Media Access for Wireless Sensor Networks” by Polastre et al.
Low Power Listening
Node goes back to sleep
If a packet is received
After a timeout
Preamble length matches channel checking period
No explicit synchronization required
Thursday’s paper!
33
LPL Limitations
Overhearing: Non-targeted receivers who sample the channel during preamble transmission have to wait until the end of the preamble to go back to sleep
Energy expenditure is a function of density as well as traffic load
Entire preamble needs to be sent before data transmission
Even though on average receiver wakes up half way through the preamble
Multiple senders need to wakeup the same receiver
34
X-MAC
Preamble contains destination ID Other receivers can
return to sleep
Strobed preamble Receiver sends ACK after
receiving short preamble
Receiver stays awake after packet reception Transmissions from
pending senders can proceed without additional preambles
35
From “X-MAC: A Short Preamble MAC Protocol for
Duty-Cycled Wireless Sensor Networks” by
Buettner et al.
36
Scheduled Listening and LPL
Scheduled listening
Advantage
Efficient transmission
Disadvantage
Synchronization overhead
Low-Power Listening
Advantage
minimizes listen cost when no traffic
Disadvantage
high costs on transmission
37
Scheduled Channel Polling (SCP-MAC)
SCP synchronizes neighbor’s channel polling time A short wake up tone wakes up receiver
It is efficient for both unicast and broadcast packets
From “Ultra-Low Duty Cycle MAC with Scheduled Channel Polling”
by Ye et al.
“You talkin’ to me?”
Inherent problem with LPL
Overhearing increases idle listening
WiFi can trigger overhearing
38 From “Design and Evaluation of a Versatile and Efficient Receiver-
Initiated Link Layer for Low-Power Wireless” by Dutta et al.
Receiver Initiated MAC
Goal:
Reduce idle overhearing in dense networks
RI-MAC
Sender does the idle listening
Receiver transmits beacons
Implementation part of the final project!
39 From “RI-MAC: A Receiver-Initiated Asynchronous Duty Cycle MAC Protocol for Dynamic
Traffic Loads in Wireless Sensor Networks” by Sun et al.
Recap: Koala network architecture
40
Soil Monitoring
Reliable data collection
Long network lifetime
Robustness to failures
Flexibility
Design outline
Motes collect measurements to local flash
Sleep most of the time (> 99%) to conserve energy
Gateway periodically wakes up the network to retrieve mote measurements
Low Power Probing
Context
Nodes sleep for weeks without radio contact
Clocks drift out of synchronization
Problem
How to resynchronize with a minimum of control packets?
No explicit acknowledgement with sender initiated wakeup
Difficult to determine when to actually stop wakeup process
41
Low Power Probing
Receiver Initiated Wakeup
Nodes send periodic beacons while sleeping
After each beacon, nodes listen for acknowledgement
Node remains awake when acknowledgement is received
Nodes that are awake acknowledges other beacons
One node initializes network wide wakeup by acknowledging the other nodes’ beacons
Result: wakeup spreads like flood wave from origin
Benefits:
Nodes can collect neighborhood information from beacons
Use neighborhood information to detect missing nodes
42
Simulcast / Concurrent Transmissions
Medium Access Control
Avoid Packet Collisions
Are all packet collisions harmful?
Simulcast / Concurrent Transmissions
Radio waves can interfere constructively and destructively
Precisely timed, identical radio packets can be decoded
Capture effect:
AGC locks onto the signal with highest SNR
Non-destructive inference:
Amplitude differences inhibits signal cancellation caused by phase differences in carrier wave
43
Simulcast / Concurrent Transmissions
Insteon
Commercial company specializing in home automation
Insteon Demo
Glossy
Rediscovered in 2011 by Ferrari et al.
TDMA based
Time synchronization for free
Flooding vs. point-to-point
LPL or LPP control packets equivalent to flooding
No hidden terminals
No routing
44
Summary
Physical Layer
Medium Access Control
Scheduled
Contention
[A-Z]-MAC
Next week:
Link Estimation and Routing
45
Schedule
46
Week 1: Introduction and Hardware
Week 2: Embedded Programming
Week 3: Medium Access Control
Week 4: Link Estimation and Tree Routing
Week 5: IP Networking
Week 6: Near Field Communication
Week 7: (seminar, no lecture)
Week 8: Energy Management
Week 9: Review and Midterm
Week 10: Time Synchronization
Week 11: Localization
Week 12: Energy Harvesting
Week 13: (seminar, no lecture)
Week 14: TBD
