TAPs: An Architecture and
Protocols for a High-Performance
Multi-hop Wireless Infrastructure
Ed Knightly
ECE/CS Departments
Rice University
http://www.ece.rice.edu/~knightly
Joint work with
V. Kanodia, A. Sabharwal, and B. Sadeghi
The Killer App is the Service




High bandwidth
High availability
– Large-scale deployment
– High reliability
– Nomadicity
Economic viability
Why?
– Broadband to the
home and public
places
– Enable new
applications
Ed Knightly
WiFi Hot Spots?

11 Mb/sec, free spectrum, inexpensive APs/NICs
Carrier’s Backbone/Internet
T1
Medium bandwidth (wire), sparse, and expensive

Why? poor economics
– High costs of wired infrastructure ($10k + $500/month)
– Pricing: U.S. $3 for 15 minutes; CH: 0.90 CHF/minute
– Dismal coverage averaging 0.6 km2 per 50 metro areas
projected by 2005
Ed Knightly
Cellular?
High availability, but slow and expensive


Cellular towers are indeed ubiquitous
– Coverage, mobility, …
High bandwidth is elusive
– Aggregate bandwidths in Mb/sec range, per-user
bandwidths at dial-up speeds
– Expensive: spectral fees and high infrastructure costs
Ed Knightly
Ad Hoc Networks?
“Free” but low availability and low bandwidth


Availability
– Problems: intermediate nodes can move, power off,
fade, DoS attack  routes break, packets are dropped,
TCP collapses, …
Low bandwidth
– Poor capacity scaling
– Unlike cellular, users consume wireless resources at
remote locations
Ed Knightly
TAPs: Multihop Wireless Infrastructure


Transit Access Points (TAPs) are APs with
– beam forming antennas
– multiple air interfaces
– enhanced MAC/scheduling/routing
protocols
Form wireless backbone with limited
wired gateways
Ed Knightly
Multihop Wireless Infrastructure





Transit Access Points (TAPs) are APs with
– beam forming antennas
– multiple air interfaces
– enhanced MAC/scheduling/routing
protocols
Form wireless backbone with limited
wired gateways
High bandwidth
– High spatial reuse and capacity scaling
– Opportunistic protocols
High availability
– Redundant paths and non-mobile infrastructure
– Deployability
Good economics
– Unlicensed spectrum, few wires, exploit WiFi components
Ed Knightly
Prototype and Testbed Deployment



FPGA implementation of enhanced opportunistic,
beamforming, multi-channel, QoS MAC
Build prototypes and deploy on Rice campus and
nearby neighborhoods
Measurement study from channel conditions to
traffic patterns
Ed Knightly
Outline

TAP architecture

OAR: an opportunistic auto-rate MAC

MOAR: multi-channel OAR

Open problems
Ed Knightly

Wireless channel is variable

Received signal:

Coherence time
superposition of different reflections,
with different delays and attenuations
40
35
30
SNR
25
20
15
10
5
Time (sec)
Ed Knightly
476
451
426
401
376
351
326
301
276
251
226
201
176
151
126
101
76
51
26
1
0
channel gain 
Motivation
time 
Opportunistic MAC Goal

channel gain 

Exploit the variations inherent in wireless channel to increase
throughput
Maintain fair temporal shares for different flows
user 1
user 2
time 
user 1
user 2

Constraint: distributed random access protocol
Ed Knightly
IEEE 802.11 Multi-rate

Support of higher transmission rates in better
channel condition
– 802.11b
available rates: 2, 5.5, 11 Mbps
– 802.11a
available rates: up to 54 Mbps

Auto Rate Fallback (ARF)
– [Monteban et al. 97]
– Use history of previous transmissions to
adaptively select future rates
Ed Knightly
Temporal vs. Throughput Fairness
Equivalent in single-rate networks
 Throughput fairness results in significant inefficiency
in multi-rate networks
Example

user 1
user 3
access point
user 2
Ed Knightly
Temporal vs. Throughput Fairness
Equivalent in single-rate networks
 Throughput fairness results in significant inefficiency
in multi-rate networks
Example

user 1
Throughput Fair
user 1
user 2
user 3
DATA
user 3
DATA
DATA
user 2
access point
Even 1 user with low transmission rate results in
a very low network throughput
Ed Knightly
Temporal vs. Throughput Fairness
Equivalent in single-rate networks
 Throughput fairness results in significant inefficiency
in multi-rate networks
Example

user 1
Temporal Fair
user 1
DATA
user 2
user 3
DATA DATA DATA DATA DATA DATA
DATA DATA DATA DATA DATA DATA
user 3
user 2
access point
Same time-shares of the channel for different
flows, also higher throughput
Ed Knightly
Opportunistic MAC
Goal
 Exploit short-time-scale variations inherent in
wireless channel to increase throughput in
wireless ad hoc networks
Issue
 Maintaining temporal share of each node
Challenge
 Channel info available only upon transmission
Ed Knightly
Opportunistic Auto Rate (OAR)

Observation: coherence time on order of multiple
packet transmission times
– measure channel quality on RTS/CTS handshake
– hold good channels for multiple transmissions

Ensure fairness by scaling number of packets
transmitted to channel quality
– # packets = Current rate / Base rate
– with random access, all flows equally likely to
access channel
OAR: High throughput, while maintaining temporal
fairness properties of single rate IEEE 802.11
Ed Knightly
RBAR Protocol
Receiver Based AutoRate (RBAR)


Receiver controls the
sender’s transmission
rate
Control messages sent
at Base Rate
Reservation Sub-Header
DATA
RTS
CTS
ACK
destination
source
Ed Knightly
OAR Protocol
Channel Condition
Protocol
BAD
MEDIUM
GOOD
Pkts
Rate
Pkts
Rate
Pkts
Rate
802.11
1
2
1
2
1
2
802.11b
1
2
1
5.5
1
11
OAR
1
2
3
5.5
5
11
OAR

Once access granted, it
is possible to send
multiple packets if the
channel is good
DATA
RTS
CTS
ACK
destination
source
Ed Knightly
Reservation Sub-Header
Performance Comparison
IEEE 802.11
Transmitter
Receiver
R
D1
C
A
RBAR
Observation II
R
D1 for the Rsame total time
D1 as singe-rate
contends
IEEE 802.11
but A
transmits
Receiver
C
C more data
A
Transmitter
OAR
OAR
Observation I
spent
in
is the same for
R
D1contention
D2 per packet
D3
RBAR and single-rate IEEE802.11
Receiver
C
A
A
A
Transmitter
Time
Ed Knightly
Performance Comparison
IEEE
802.11
Observation
III
R
D1 channels for multiple
holds
high-quality
transmissions
Receiver
C
A
Transmitter
OAR
RBAR
Transmitter
R
D1
C
Receiver
R
A
D1
C
A
OAR
Transmitter
Receiver
Ed Knightly
R
D1
C
D2
A
D3
A
A
Analytical Model

Challenge: MAC and channel are random processes with
memory

Model relates physical-layer characteristics to MAC throughput:
– Time spent in contention

Markov model of modified IEEE 802.11
– Average transmission rate


Due to channel distribution
Comparative model b/t multi-rate OAR and tractable systems
– TIME: OAR contends as often as single-rate IEEE 802.11
with increased data per contention
– PACKETS: OAR reduces packet transmission time via percontention rate adaptation
Ed Knightly
Simulation Results Under Ricean Fading
Packets Transmitted
with 11 Mbps (%)
Protocol
OAR
80%
RBAR
65%
Nodes



OAR has 42% to 56% gain over RBAR
Increase in gain as number of flows increases
Model predicts OAR & RBAR throughput to within 7% accuracy
Ed Knightly
Outline

TAP architecture

OAR: an opportunistic auto-rate MAC

MOAR: multi-channel OAR

Open problems
Ed Knightly
Multi-Channel Problem Formulation

Observe: for two MUs, quality of different channels
can have low correlation if
channel separation >> coherence bandwidth
– Example: at
2.4 GHz WiFi,
5 vs. 1-3 MHz
– Figure for
Ricean, K=4
Ed Knightly
Challenge

Ideal protocol is simple: select the best channel
at the instant of transmission

In practice, channel qualities are unknown
a priori
– Must first transmit and measure

Cost of measuring channels must be balanced
with benefits of finding good ones
Ed Knightly
MOAR Protocol Sketch

Measure channel SNR at RTS/CTS handshake

If channel quality is high (above an SNR
threshold), transmit via OAR

If channel quality is poor, skip to a new channel
– next channel piggybacked in CTS

Design optimal stopping rule for skipping
– stop when throughput gain of skipping to a
better channel is outweighed by overhead

Ensure fairness
Ed Knightly
Optimal Stopping Rule Formulation

Let Xn denote the SNR of the nth measured channel

Let c denote the cost (in time) of measuring the channel

After observing Xn transmit or measure again?
– cannot go back to previous channel (coherence time)

The reward for the nth selection is Xn-nc
– after scaling SNR to rate and then to time

Objective: maximize the expected reward

In a class of stopping rule problems (without recall)
Ed Knightly
Optimal Stopping Time

Let V* denote the expected return from the optimal
stopping rule

Suppose pay c and observe X1= x1

If continue, x1 is lost and c is paid
– continuing, can obtain return V*, but not more
– start afresh

Optimal rule is threshold based
– If xn < V*, continue; if xn > V* stop
– N* = min{n  1: Xn  V*}
Ed Knightly
Calculating the Stopping Threshold V*


V* = E max(X1,V*) – c


V*
( x  V *)dF ( x)  c
– F(x) represents the SNR distribution

Compute V*
– channel model and parameters (ex. K, d)
– system’s rate-SNR thresholds (ex. 1, 2, 5.5, 11)
Ed Knightly
MOAR Throughput Gains

Gains of 40%-60% increasing with K and SNR
variance
Ricean
parameter
K = 0 is no
line-of-sight
signal
Ed Knightly
Effect of Node Distance


Greatest help when far away
Non-monotonic due to rate-SNR thresholds
Ed Knightly
Random Topologies

Nodes are uniform-randomly
placed in a 250m circle

“Optimal Skipping” cheats:
looks at all channels (with no
cost) and jumps to the best

Observe
– MOAR extracts most
available gain
– close-by nodes detract
from average gain
Ed Knightly
Outline

TAP architecture

OAR: an opportunistic auto-rate MAC

MOAR: multi-channel OAR

Open problems
Ed Knightly
DoS Resilience and Security

Old methodology
– Design a network protocol
– Optimize for performance
– Discover DoS/Security holes

Ex. Route query floods
– Patch one-by-one

Challenge
– DoS-resilience and security as the foundation of
network protocols
– Recognize these issues are as important as
performance
Ed Knightly
TAP Media Access and Scheduling

Challenge: distributed scheduling
– Others’ channel states, priority, & backlog condition unknown


Ex. TAP A’s best recv’r may be transmitting elsewhere
Ex. Traffic to be recv’d may be higher priority than that to be sent
– Traffic and system dynamics preclude scheduled cycles
– Modulate aggressiveness according to overheard information
Ed Knightly
Multi-Destination Routing/Scheduling



Most data sources or sinks at a wire
Routing protocols for any wire abstraction
Scheduling
– At fast time scales, which path is best (channels,
contention, …) now?
– Can delay/throughput gains be realized despite
TCP?
Ed Knightly
Distributed Traffic Control

Distributed resource management: how to throttle
flows to their system-wide fair rate?
– Throttle traffic “near-the-wire” to ensure fairness
and high spatial reuse
– TCP cannot achieve it (too slow and RTT biased)
– Incorporate channel conditions as well as traffic
demands
Ed Knightly
Capacity Driven Protocol Design
Protocol Driven Capacity Analysis



Traditional view of network capacity assumes zero protocol
overhead (no routing overhead, contention, etc.)
Protocols themselves require capacity
A new holistic system view: “the network is the channel”
– Incorporate overhead in discovering/measuring the resource
– Explore capacity limits under real-world protocols
Ed Knightly
Problem: Multiple APs/TAPs/…
within Radio Range


PHY Interference has disproportionate throughput degradation
at MAC layer
Interference can lead to severe scaling limitations and
starvation (worse than zero-sum game)
Ed Knightly
Summary

Transit Access Points
– WiFi “footprint” is dismal
– Removing wires is the key for economic viability

Opportunistic Scheduling (OAR/MOAR)
– Exploit time and frequency diversity

Challenges
– Multi-hop wireless architectures
– Distributed control
– Scalable protocols
Ed Knightly