RadioActive Networks: Robust Wireless Communications

1
RadioActive NetworksRobust Wireless
Communications

• John Ankcorn
• Stephen Garland
• John Guttag
• Matt Welborn (XtremeSpectrum)

Networks and Mobile Systems Research Group MIT
Laboratory for Computer Science nms.lcs.mit.edu
2
The wireless network challenge

• Things change, even in wired networks
• Applications, people, the world, ...
• Active networks address the challenge of change

Change is faster, less predictable in wireless
networks People move Condition of communications
channel changes RadioActive networks address the
larger challenge
3
Digital wireless communication
0 0 1 1 0 1 1 1 0 0
0 0 1 1 1 1 1 1 0 0
4
Problems with conventional networks

• They treat most application as having similar
  needs
• Bandwidth, latency, error rate
• Security, error recovery

They target the worst (or average) case But
there are many different ways to be bad Which
rarely all happen at once
Consequently, conventional static designs
... Waste resources almost all the time Perform
unnecessarily poorly much of the time
5
Meeting the wireless challenge

• Adapt dynamically to the existing case
• Current channel conditions
• Current application
• Some kinds of adaptation occur today
• Dynamic channel allocation
• Adaptive power control (for transmission)

RadioActive networks promise much more
6
RadioActive networks

• Virtual or software radios
• Provide adaptive physical layer
• Two examples using SpectrumWare
• Dynamically changing what is transmitted
• Dynamically changing behavior of receiver
• Active networks
• Provide control channel
• John Guttag talked about this before

7
An adaptive wireless network interface
OSI network layers
Virtual radio layers
Bytes
Link framing
Data link
Bits
Channel coding
Bits
Line coding
Symbols
Modulation
Discrete signal
Multiple access
Discrete signal
Physical
A/D conversion
Continuous signal
Freq. conversion
Continuous signal
8
Ideal software radio

• Not exactly what we have, but close enough for
  today

Wideband antenna
Wideband analog/digital converter
Processor and Memory
A/D
Amplifiers
D/A
9
An admission

• No live demo today
• We have done table-top demos at GLOMO meetings
• They involve considerable hardware, preparation
• Not appropriate for this format
• Will instead show videos, screen shots

10
SpectrumWare radio overview (video)
11
Other software applications
Connect people Versatile patch panel Like the
good old days
Use for healthcare Connect
ambulance/hospital Universal instruments Interchan
geable probes Ultrasound, EKG, EEG, A major
activity now at MIT

• Connect computers
• Universal wireless
• network interface

12
Adaptive transmission and reception

• Transmission
• Change what is transmitted
• Location and width of channel
• Coding of symbols
• Reception
• Change algorithm for processing received signal
• E.g., to reduce power consumption
• Out-of-band or in-band signaling to coordinate
  change
• Via ActiveNet technology

13
Transmission

• Modulation represents information on physical
  channel
• Variation in amplitude and/or frequency/phase
• Analog information AM, FM
• Digital information (symbols) PAM, QAM, ...
• Over a clear channel
• Symbols can be close together
• Allows sending more data in same width band
• Over a noisy channel
• Symbols can be spread apart
• Reduces bit error rate

14
Modulation 8-PSK with high SNR
15
Modulation 8-PSK with low SNR
16
Modulation QPSK with low SNR
17
Reception

• Two basic issues
• Channel separation (eliminate interfering
  signals)
• Demodulation (symbol detection)

Data Bits
Data Bits
Shared channel
Demodulation
Channel separation
Modulation
Receiver
18
Reception channel separation

• Desirable kinds of adaptation
• To interference (FDMA) and noise
• To application requirements for output SNR
• Challenge
• Make computation depend on output sample rate
• Not on the higher input sample rate
• Example Wideband cellular phone receiver (AMPS)
• Band width Sample rate
• AMPS band 10MHz 25M/sec
• Single channel 30kHz 60K/sec

19
Channel separation overview
20
Bandwidth reduction

• Time duration of input (output channel width)-1

21
Reception symbol detection
To decide whether transmitted bit was 1 or
0 1) Compute filter output
22
Adaptive symbol detection
Receiver uses Quality of channel Desired error
rate To control Accuracy Power consumption Bit
rate (with protocol)
10
0
-10
Threshold
10
20
30
N
Terms in sum
10
0
Software optimization Sort samples!
-10
10
20
30
N
23
How many terms to compute?

• Calibrate
• For n 1..N
• Plot bit-error rate vs. ?
• To achieve a given BER
• Pick n, ?

BER
n 3
10-1
10-3
n 4
10-5
n 10
?
10-7
Test after n steps If sum lt ?, test again
after N steps
24
Wrapping up

• Virtual radios allow
• Faster innovation
• Easier adaptation
• Graceful degradation
• Cross layer optimization
• Research areas for RadioActive networks
• New signal processing algorithms
• Network protocols that exploit flexibility
• End-to-end analysis of soft physical layers