Page 1 Robust Ultra High Frequency (UHF) Satellite Communications Protocol for UUVs SBIR Topic #:...

Robust Ultra High Frequency (UHF) Satellite Communications

Protocol for UUVs

SBIR Topic #: N02-019Contract #: N66604-02-C-4577

Deliverable Item 0001ACPresentation on Phase I Work

Wavix, Incorporated27 January 2003

Phase-I Objectives

• Understand the maritime noise environment

• Characterize RF (UHF) communication impediments

• Identify mitigating (low-level) protocol techniques

• Recommend development paths to follow

The Challenge

Maritime Noise ConcernsOriginal List

• Signal fading from wave motions rocking UUV

• High seas coating antenna or changing local RF propagation characteristics

• Shadowing by waves• Surface wave reflections causing multi-path

interference• Changing atmospheric conditions in LOS• Nature of channel noise: high BER?

Maritime Noise ConcernsNon-Obstacles I

• Antenna wash-over– Choose appropriate non-conductive housing– Water has small attenuation at UHF

• Ground-plane issues– Appropriate choice of antenna:

• Quadrifilar Helix has small sensitivity to ground plane

• Helical antenna is very sensitive to ground plane

Maritime Noise ConcernsNon-Obstacles II

• Water aerosols– Water has little attenuation in UHF band– (Atmospheric water effects become

serious above ~2-4 GHz)

• UUV rocking– Motions are too small to disrupt signals– Motions are too slow for Doppler effects

Characteristic Length Scales

• Wave Heights: 2 m• RF Wavelength: 1 m (@ 300

MHz)• Surface Roughness: 0.1 m

Characteristic Time Scales• Wave times: ~5-10 s• SATCOM frames:

– 8.96 s, made of 1024 blocks [5 kHz waveform]– 1.3866 s, made of 0.052 ms time chips [25 kHz

waveform]• Geo propagation time: 500 ms• SATCOM blocks:

– 8.75 ms blocks in 5 kHz waveform; variable number/channel

– 0.052 ms “time chips” in 25 kHz waveform; channel format varies but order of 100 might be typical

• ATM packet size: ~2.5 ms – IP packets are 4 to 10 times larger, but variable

• Symbol size: ~0.05 ms (@ 19.6 bits/second)

Maritime Noise ConcernsObstacles

• Small look angle to satellite

• Wave obscuring• Ocean-surface roughness• Nature of the RF channel

Satellite Elevation vs. Latitude

Latitudes = Elevations

(for G = 0!)

20 = 82

30 = 71

40 = 56

50* = 40

60 = 25

70 = 12

(* US/Canadian border,

English Channel, etc.)

Waves Obscuring LOS

• Fading by attenuation at small look angles; some multipath at large look angles

• Operations in northern latitudes: near-horizon viewing of SATCOM satellites

• Operations in sea-state four: wave heights of 1.25 to 2.5 meters (average = 1 PI)

• Depth of fades: ~7 dB @ 1 GHz• Wave period: 5 to 10 seconds • Time scale of fades: ~1-3 seconds

Ocean Surface Roughness

• Fading effects from multi-path interference

• Length scales of ~0.1 meter mean less interaction with UHF signals

• Most serious at very small look angles

Nature of the RF Channel

• The RF channel is principally a fading channel (Rayleigh channel, or channel with memory), and not a noisy channel (AWGN channel, or memoryless channel)

• BER has limited utility• Memory channels are much more

difficult to simulate (Markov chains)• Much less research exists for fading

channels

Protocol Strategy

Look at low-level protocol strategies that will have the greatest utility in

mitigating the predominantly fading maritime-communications channel.

Illustration courtesy Catherine Werst

BER vs. S/N

0 1 2 3 4 5 6 7 8 9 10 11 12 13 1410

-6

10-5

10-4

10-3

10-2

10-1

100

BE

R

QPSK(BPSK)

Non-coherent FSK

Eb/N

o (dB)

Tradeoffs Discussed

• UUV System Configuration

• Satellite Access Methods• Physical Layer:

Modulation• Physical Layer: Coding• Link-Layer Directions &

Recommendations

UUV System Configuration

• Antenna– Type of antenna– Housing for antenna– Antenna on mast– Antenna diversity

• Power– A system issue that

affects RF performance

Satellite Access Methods

• Single-User Channels• Frequency Division: FDMA• Code Division: CDMA

– Spread-spectrum approach would be challenging in satellite environment because of dynamic signal balancing

Time Division: TDMA [& GSM] – Provides adequate multi-use capabilities– Mature technology– Compatible with SATCOM, although different

specifications may be desirable

System-Level Schematic

Bit-StreamTo Packet

ChannelEncoding

Mod.& Xmit

Packet toBit-Stream

ChannelDecoding

Rec. & De-Mod.

Air Link

PhysicalLayer

LinkLayer

Bits rate vs. Symbol rate

Symbol Transmissions (baud)

Bi-state:1 bit/symbol

4-state:2 bit/symbol

Physical Layer: Modulation

I

Q

I

Q

01 00

11

01

10

I

Q

I

BPSK QPSK

“M-ary” PSK

Trellis-coded

Modulation Tradeoff

The result from information theory limits

information rate/bandwidth(i.e., baud rate), but not bits/second,

which can be increased with a compensating increase in transmit

power.Higher-order modulation techniques

are useful for bandwidth-limited applications, but we recommend

QPSK .

Channel Coding &Forward Error Correction

• Channel coding describes the process by which a logical bit stream gets turned into a modulated signal suitable for transmission of the desired information.

• Forward Error Correction (FEC) describes coding techniques that encode the bit stream so that errors in the received bit stream can be corrected.

• Coding & FEC all take place in the lowest layer of the protocol stack, transforming a bit stream into its most desirable form for modulating the carrier for transmission.

• The benefit of an FEC technique is described as its coding gain.

• FEC is not the same as error detection (e.g., CRC bits).• Not all channel coding is FEC (e.g., NRZI).

FEC Techniques & Tradeoffs• Block Codes, which operate on blocks of

symbols, are generally better on block errors– Hamming Codes– BCH codes– Reed-Solomon Coding

• Convolutional Codes, which operate on the stream of bits, are generally better on bit errors– Convolutional Codes – Viterbi Decoding – Turbo Codes (Turbo anything is very hot)– Turbo Product Codes

• Coding gains are generally 2 – 3 dB

Reed-Solomon Coding I

• RS coding operates on a block of symbols, not modifying it but adding additional bits to the stream that can correct errors in the block.

• RS(n,k) – n encoded symbols; k message symbols– t = (n – k)/2 symbols can be corrected

• The degree of error correction depends on the number of added bits (2t).

• Adding bits increases bandwidth overhead.

Reed-Solomon Coding II• In principal: Modern theoretical treatments of Reed-

Solomon coding use Galois group theory [GF(28)]!• In brief: combinations of correcting bits can

efficiently identify erroneous combinations of information bits

• In practice: The algorithms are widely available as firmware.

• NASA specifies RS(255,239) or RS(255, 223) for deep-space missions.

We recommend RS(255,239) for its generally good performance and easy availability.

Block Code Performance IG

rap

h b

y K

. Aza

det

IEEE 8

02

.3 H

igh

-Sp

eed

Stu

dy

Gro

up

Ple

nary

meetin

g, M

on

treal

July

19

99

Block Code Performance IIG

rap

h b

y K

. Aza

det

IEEE 8

02

.3 H

igh

-Sp

eed

Stu

dy

Gro

up

Ple

nary

meetin

g, M

on

treal

July

19

99

Convolutional Encoding Concept

1 2 3 4 5 … n

Constraint length (n); rate (1/2); puncturing.

+

+

Xmit a

Xmit b

Convolutional Encoding Characteristics

• Operational choices:– Constraint length (length of shift register)– Generating polynomials (useful constraint lengths

are generally known to have optimal choices)– Symbol rate (determined by number of adders)– Puncturing (to increase the symbol rate at the cost of

decoding difficulty)• Our recommendation, again, is based on

simplicity, easy availability and, in this case, SATCOM compatibility.

• We recommend: Constraint length 7 Rate to be determined, but ½ and ¾ are common

Viterbi Decoding(with Soft-Decoding Information)

•In principal: Maximum-Likelihood Estimation•In practice: Algorithms available in firmware Soft-decoding uses signal strength to assist in decision making, for a gain of ~2 dB.

Rec. bitstream

etc.

11

10

01

00

Concatenation Concept

If one coding scheme is good,wouldn’t two be better?

In fact, Reed-Solomon and Convolutional Encoding are

complementary and commonly used together.

The Importance of Interleaving

• Interleaving effectively transforms block errors into bit errors.

• Interleaving is neither error-correcting or error-detecting; it is an error avoidance technique.

• There is no coding gain associated with interleaving

• The tradeoff in choosing the size of the interleaver is between time scale of bit dispersal and tolerable delay times in transmission.

x1

x2

x3

x4

x1’ …

r1 …

r1’

r2’

r3’

r4’

Recommended Concatenation

Reed-Solomon Outer Code – RS(255, 239)

Large-Order Interleaving– Scale to be determined by communication

constraints, but scaled to be effective against fading from waves

Convolutional Inner Code– Constraint length 7– Rate ½ to ¾– Viterbi decoder with soft-decision

• Coding gain ~5 dB => • BER reduction from ~10-3 to 10-9

Physical Layer Diagram

(Link-Layer Packets)

Reed-Solomon Encoding, RS(255,238)

Large-Scale Bit Interleaving

Convolutional EncodingL=7, R=1/2 or 3/4

NRZI Encoding

QPSK modulation

Transmit

Receive

(Link-Layer Packets)

Reed-Solomon Decoding, RS(255,238)

Large-Scale Bit De-Interleaving

Viterbi Decodingwith Soft-Decision

NRZI Decoding

Quadrature Demodulation

Antenna Diversity

Link Layer Recommendations

• Minimize Automatic Retransmit Request (ARQ)– Reserve it for higher-level protocols

• Minimize handshaking• Recognize transmission delay times• Build a delay-tolerant network

Metaframing

• A concept we introduced in the proposal, but didn’t develop in Phase I

• Not SATCOM compatible, but similar in framework.

• Requires:– Good receiver timing and clock

synchronization– Reconsideration of guard times– Data backcapture to achieve gains

Hierarchy of Recommendations

• Easy– Implement within UHF SATCOM

TDMA/DAMA capability• Medium

– Protocols beyond SATCOM capabilities– Possibly implemented over dedicated

SATCOM channels– May require new radios– May affect UUV system design

• Hard– Not compatible with existing SATCOM

specifications

Easy Recommendations

• Use Quadrifilar Helix Antenna• Use existing TDMA/DAMA satellite

access • Use existing QPSK Modulation• Use existing convolutional inner

code– length 7, rate ½ or ¾

• Add additional interleaving and RS(255,239) outer code if possible

Medium Recommendations

• Implement antenna diversity• Implement in custom-designed radio:

– Full concatenated channel coding:• Reed-Solomon outer code: RS(255, 238)• Interleaving (depth to be determined)• Convolutional inner code: length 7, rate ½ or ¾

– Viterbi decoder with soft-decoding

• Define delay-tolerant link-layer through transport-layer protocols.

• Possibly redesign TDMA specifications (over dedicated SATCOM channel)

Hard Recommendation

Build a new, non-geostationary satellite system that will give

significantly better coverage over the oceans and ease the RF

communication channel’s susceptibility to fading because of high seas and small look angles.

Page 1 Robust Ultra High Frequency (UHF) Satellite Communications Protocol for UUVs SBIR Topic #:...

Documents

Transcript of Page 1 Robust Ultra High Frequency (UHF) Satellite Communications Protocol for UUVs SBIR Topic #:...