p14251 underwater acoustic communication -...

46
Greg Davis Scott Hambleton Jon Holton Chris Johnson Chris Monfredo 12/10/13 Rochester Institute of Technology 1 P14251 Underwater Acoustic Communication

Upload: phamkhuong

Post on 29-Mar-2018

216 views

Category:

Documents


1 download

TRANSCRIPT

Greg DavisScott HambletonJon HoltonChris JohnsonChris Monfredo

12/10/13 Rochester Institute of Technology 1

P14251

Underwater Acoustic Communication

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 2

AgendaDesign Updates/Proof of ConceptSoftware ComponentsDrawings (EE)BOM (EE)Drawings (ME)BOM (ME)Final Cost AnalysisFinal Risk AnalysisTest PlansMSD II Schedules

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 3

Frequency Shift Keying:

Produces four different frequencies corresponding to 00, 01, 10, 11 SD Simple modulation and demodulation schemes and circuitry Quick and efficient set up and delivery times Schedules

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 4

Frequency Shift Keying:

Modulation:Theory-produce four different frequencies that each represent a different two bit symbol. These frequencies will be gray coded such that only a maximum of one bit error can occur per demodulation error.

• After compression and redundancy, we need to send 9768 kbps to meet our 15 kbps customer requirement. This indicates that the lowest possible frequency

that can be used is 14.4 kHz.

29 kHz = 0034 kHz = 01 39 kHz = 1144 kHz = 10

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 5

Frequency Shift Keying:

Modulation:• To produce the four needed frequencies, properties of a square wave can be

exploited. The main property being used is that in frequency, a square wave, is sync function. A sync function has a large amplitude at the center frequency and multiple harmonics that theoretically occur every 3*fc

• Using a sharp band pass filter will allow us to keep the large portion of the signal that occurs at the center frequency while removingall of the other harmonics

• To generate the needed frequency response, an LCband pass filter will be used

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 6

Frequency Shift Keying:

Modulation:• The LC band pass filter has a center frequency at 1/sqrt(LC), and a bandwidth of

1/(C*RL). Since the output of our filter will be fed into an amplifier with theoretical infinite input resistance, our bandwidth will be very small; which is desirable.

• By adjusting the input resistance, we can also adjust the sharpness of each filter• Each filter will be fed by a clock pin coming from the RPi, that will provided the

square wave at each of the four desired frequencies• The output of the four band pass filters will be fed into a 4:1 mux whose output

will be controlled via an enable and two addressing pins

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 7

Frequency Shift Keying:

Modulation:

-80

-60

-40

-20

0

20

Magnitu

de (

dB

)

104

105

106

-90

-45

0

45

90

Phase (

deg)

Bode Diagram

Frequency (rad/s)

-100

-80

-60

-40

-20

0

20

Magnitu

de (

dB

)

104

105

106

107

-90

-45

0

45

90

Phase (

deg)

Bode Diagram

Frequency (rad/s)

-100

-80

-60

-40

-20

0

20

Magnitu

de (

dB

)

104

105

106

107

-360

-270

-180

-90

0

Phase (

deg)

Bode Diagram

Frequency (rad/s)

-100

-80

-60

-40

-20

0

Magnitu

de (

dB

)

104

105

106

107

-360

-270

-180

-90

0

Phase (

deg)

Bode Diagram

Frequency (rad/s)

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 8

Frequency Shift Keying:

Demodulation:• To demodulate we are looking to decipher the four different frequencies in order

to obtain the original binary message• This will be done by taking the incoming signal from the hydrophone, band

passing around our frequency range, and then passing the analog signal through an analog to digital converter. The output of the ADC will then be fed back to a pin on the RPi.

• The DFT will then be performed on the incoming signal in order to determine the largest frequency component. Based on the frequency and the amplitude at this frequency, the incoming message will be decoded.

• The DFT will be performed using the FFTW C programming package developed by MIT

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 9

Frequency Shift Keying:

Demodulation:• The FFTW package allows the user to do one or multiple dimension FFT’s with a

very simple to use library• This package is also optimized for multithreading and also gives the capability to

determine the most optimum way in which to take the FFT of the current signal.#include <fftw3.h>{fftw_complex *in, *out;fftw_plan p;...in = (fftw_complex*) fftw_malloc(sizeof(fftw_complex) * N);out = (fftw_complex*) fftw_malloc(sizeof(fftw_complex) * N);p = fftw_plan_dft_1d(N, in, out, FFTW_FORWARD, FFTW_ESTIMATE);...fftw_execute(p); /* repeat as needed */...fftw_destroy_plan(p);fftw_free(in); fftw_free(out);

}

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 10

Frequency Shift Keying:

Demodulation:periods: 3

FS: 2.25 MHz

-6 -4 -2 0 2 4 6 8 10 12

x 104

95

96

97

98

99

100

101

102

-4 -2 0 2 4 6 8 10 12

x 104

83

84

85

86

87

88

89

-4 -2 0 2 4 6 8 10 12

x 104

71

72

73

74

75

76

77

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 11

Dish Concept:

• Use parabolic collector around hydrophone.

• Center hydrophone at focal point.

• Increase gain from transmitter. Parabolic Antenna ConceptRadartutorial .eu

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 12

Initial Dish Design:

• Mounts on top of housing

• Made from plastic or metal

• 12” Diameter to increase gain

11/26/13 Rochester Institute of Technology 13

P14251

Underwater Acoustic Communication

CE Overview- Software Architecture- Control unit flowchart and pseudo-code- Error Detection/Correction- Framing Information- Data Rate Analysis

11/26/13 Rochester Institute of Technology 14

P14251

Underwater Acoustic Communication

Software Architecture PC_Interface

+receiveMessage(): string

+sendMessage(string msg)

Tx_PC

+transmit(string msg)

Rx_PC

- empty : boolean

+getMessage() : string

Rx_Signal

-channel_empty : boolean

+readFrame() : bit[]

- dsp(bit[] bits)

Controller

-busy : boolean

-waitingOn : string

+main() : int

Tx_Signal

+transmitFrame(bit[] frame)

Compressor

+compressMsg(string message) : bit[]

+decompressMsg(bit[] data) : string

Encryptor

+encrypt(bit[]* data)

+decrypt(bit[]* data)

DataPacker

+createFrame(bit[]* data)

+getFrameType(bit[] frame) : string

+isRepeatFrame(bit[] frame) : boolean

+extractData(bit[] frame) : bit[]

+createControlFrame(string type) : bit[]

ErrorHandler

+hasErrors(bit[] frame) : boolean

+correctErrors(bit[]* frame)

11/26/13 Rochester Institute of Technology 15

P14251

Underwater Acoustic Communication

Control Unit Pseudo-Code

11/26/13 Rochester Institute of Technology 16

P14251

Underwater Acoustic Communication

Control Unit Pseudo-Code

11/26/13 Rochester Institute of Technology 17

P14251

Underwater Acoustic Communication

Control Unit Pseudo-Code

11/26/13 Rochester Institute of Technology 18

P14251

Underwater Acoustic Communication

Error Detection and Correction

- Hybrid Scheme: ECC and ARQ- EEC Implementation: BCH vs. Reed Solomon- BCH is easier to implement, but requires a much larger

amount of redundancy- Reed-Solomon is more complex, but overall much

better and requires only 20% redundancy (to correct 10% of errors)

11/26/13 Rochester Institute of Technology 19

P14251

Underwater Acoustic Communication

Error Detection and Correction

- Reed Solomon encodes k symbols into n codewords- n – k = 2t - Errors are corrected at the symbol level. If a symbol

has 4 bits and all of them are wrong, it only counts as one error.

- Can vary the code word size with the number of bits per symbol

- RS(255, 212) with 4 bits/symbol, RS(1023, 853) with 1 bit/symbol, etc.

11/26/13 Rochester Institute of Technology 20

P14251

Underwater Acoustic Communication

Error Detection and Correction

- Encoding: LFSR based implementation is extremely simple and fast. Will translate well to C

11/26/13 Rochester Institute of Technology 21

P14251

Underwater Acoustic Communication

Error Detection and Correction

- Decoding is more complex, but efficient algorithms exist that will help significantly

11/26/13 Rochester Institute of Technology 22

P14251

Underwater Acoustic Communication

Frame Sentinels- Sentinels are a unique pattern of bits that signify the

start and end of the framei.e. 01110 –data– 01110

- When preparing the frame for transmission, if the pattern appears anywhere in the data, bit stuffing is used to eliminate it (i.e. 01110 -> 011010)

- Commonly used sentinel which we’ll use is 01111110Frame Header- 1 bit to signify the type of frame: control or message- 1 bit that gets flipped each time a new frame is sent

11/26/13 Rochester Institute of Technology 23

P14251

Underwater Acoustic Communication

Control frame formats- Since it’s especially important to interpret control

frames correctly, 4 bits are used to display the unique patterns

- 0000 – Request to Send (RTS)- 0110 – Clear to Send (CTS)- 1001 – Acknowledgement (ACK)- 1111 – Done (Signifies that all frames have been sent)

- These 4 bits + the 2 header bits can be encoded with an RS(7, 3) code (2 bits per symbol)

11/26/13 Rochester Institute of Technology 24

P14251

Underwater Acoustic Communication

Frame Sizes- Control Frame: 16 (sentinels) + 14 (data) = 30 bits- Message Frame: 16(sentinels) + X+2 (message) = 18+X bits

Propagation delay (20ms for 30m distance) limits the number of frames that can be sent.

For 15kb/s, the maximum number of message frames =9 frames, each containing 1k encoded information bits

Increase X to ~1023 bits for Reed-Solomon Encoding

11/26/13 Rochester Institute of Technology 25

P14251

Underwater Acoustic Communication

Data Rate Analysis- fc = 37kHz, tc = 81μs (3 periods per 2 bits)- Frames that get transmitted for a message:

- Send RTS – 30 bits- Receive CTS – 30 bits- Send all message frames – 1042 bits/frame- Receive all acknowledgements – 30 bits/frame- Send Done – 30 bits- Receive Acknowledgement – 30 bits

- Total transferred bits = 9768 bits- Total transfer time = 0.836s (about 170ms left for code)

11/26/13 Rochester Institute of Technology 26

P14251

Underwater Acoustic Communication

Data Rate Analysis – Code Overhead- From SSDR: Compression time is expected to be negligible- Encryption time is a non-factor- FFTW benchmarking info suggests

very quick performance times- Error Encoding time is negligible- Error Decoding time may take a bit longer, but millions of

clock cycles are available to work with.* We may be able to further lower the number of message frames to eliminate propagation delays. This adds slightly more complexity to error handling.

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 27

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 28

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 29

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 30

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 31

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 32

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 33

Mechanical DrawingsAssemblySheet Metal HousingMounting PlateFront and Back covers

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 34

Assembly Drawing

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 35

Sheet Metal Housing

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 36

Mounting Plate

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 37

Back Cover

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 38

Front Cover

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 39

Mechanical BOM

Total Mechanical cost: $113 + Housings and Dishes

Description Distributor D. Part Number QTY per housing Price Spares QTY Total Price Extended

Nylon Liquid-Tight Cord Grip .16" to 0.31" McMaster Carr 69915K51 2 $2.82 1 5 $14.10

8-32 CLS Self Clinching Nuts With Housing - 44 0 88 $0.00

50x 8-32 Stainless Pan Head Machine Screws 7/32" long McMaster Carr 91735A193 2 $7.86 0 4 $31.44

50x 8-32 Steel Hex Nuts McMaster Carr 90480A009 1 $1.49 0 2 $2.98

8-32 1 1/2" Long Hex Standoffs McMaster Carr 93620A945 4 $1.74 0 8 $13.92

Internal Mounting Plates 6"x6" 1/8" thick McMaster Carr 88685K1 1 $6.20 0 2 $12.40

6061 Al 6"x36" 1/8" thick McMaster Carr 8975K923 1 $17.19 0 1 $17.19

316 Stainless Sheet Metal Housing ? ? 1 0 2 $0.00

Santoprene Rubber Gasket 12"x24" 1/32" thick McMaster Carr 86215K21 - $10.45 0 1 $10.45

5 ft Adhesive Backed Velcro 1" wide McMaster Carr 9273K11 - $5.78 0 1 $5.78

100x M2.5 Steel Screws 10mm length McMaster Carr 92005A071 1 $4.13 0 1 $4.13

50x M2.5 Plain Steel Nuts McMaster Carr 90592A006 1 $1.04 0 1 $1.04

Speaker Dish 1 2 $0.00

Hydrophone Dish 1 2 $0.00

Speaker Mount 1 2 $0.00

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 40

B117-11 Spray Test Results:Test Material: Specimen # Initial Mass (grams): Final Mass (grams): Mass Change (grams):

316SS 1 64.2 64.2 0

316SS 2 70.3 70.3 0

316SS 3 62.7 62.7 0

316SS 4 66.8 66.8 0

316SS 5 62.3 62.3 0

ABS Plastic 1 8.8 8.8 0

ABS Plastic 2 8.2 8.2 0

ABS Plastic 3 8.5 8.5 0

ABS Plastic 4 8.6 8.6 0

ABS Plastic 5 8.4 8.4 0

Naval Brass 1 70.3 70.3 0

Naval Brass 2 71.5 71.5 0

Naval Brass 3 68.9 68.9 0

Naval Brass 4 73 73 0

Naval Brass 5 66.8 66.8 0

Alum 6061 1 22.1 22 0

Alum 6061 2 20.4 20.4 0.1

Alum 6061 3 22.6 22.6 0

Alum6061 4 23.8 23.7 0.1

Alum 6061 5 20.8 20.6 0.2

• 6061 Aluminum preformed worst.

• Naval Brass, ABS Plastic, and 316SS all had no noticeable effects of corrosion

• 316SS Best choice for cost and machinability.

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 41

Cost AnalysisMSD I Test Cost: $100Mechanical Component Cost: $113 + HousingsElectrical Component Cost: $1105 + PCBsTotal Budget: $1750$432 remainder for shipping and emergencies

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 42

Electrical Test PlansPower Converters

• <5% Ripple at 1A load for 5V• <1% Ripple at 1A load for 3.3V

Power Amplifier• Adjustable Gain• Find resistance for 10W

AGC• 1V Amplitude for any input

LC Filters• Adjust values

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 43

Software Test Plans- Unit test all code- Device testing : make sure code actually runs on the

Raspberry Pi- Wired device testing : make sure both devices can

communicate via wired connection before attempting wireless

- Turn functionalities on and off to see that they make a difference

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 44

Mechanical Test Plans

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 45

Risk AnalysisID Problem Likelyhood Damage Importance Mitigation Owners

1 The housing isn't watertight 2 3 6 Test waterproofing and test the empty housing SH, GD

2 Short circuit 1 3 3 CJ, CM

3 Damage due to mis-handling parts 1 2 2 Team must be c areful with components All

4 Loss of carrier frequency 1 3 3 Have a robust communication scheme in place CJ, CM

5 Power loss 1 1 1 CM

6 Power Surge 1 3 3 Surge Protection CM

7 Data loss 2 3 6 Have tested error correction/detection JH

8 Overheat microprocessor 1 3 3 Use efficient code and have thermal management JH, GD

9 Corrosion breach 1 3 3 Galvanize or use corrosion resistant materials SH, GD

10 Speaker doesn't work 1 1 1 Has been found to work, repeatably in a lab environment CJ, CM

11 Ordered components do not match specs 2 3 6 Order from reputable sources with return policies All

12 Demodulation Chip doesn't work 2 3 6 Do DSP on the R-Pi. CJ (JH)

13 Bandpass Filter doesn't work 1 1 1 CJ

14 Mode Choke doesn't work 1 1 1 CJ

15 Sheet metal housing is too expensive 1 3 3 Highly unlikely given current financial status of the project GD, SH

16 PCB is too expensive 1 2 2 Use a perf board and hand solder CM

17 Speaker/Hydrophone dish doesn't work 2 2 4 We check functionality first, then prioritize the dish SH, GD

18 Power Converteres can have too much ripple 1 2 2 Scale C and L components or use higher current capacity converter CM

19 AGC doesn't work 2 1 2 There are plenty of AGC chips available CM

20 Power amplifier doesn't work 2 2 4 TI has technical support for our power converter CM

21 Noise on and/or electrically charged box 1 3 3 Ground the box CM

22 Discharging the batteries too low 2 3 6 Implement a battery level indicator CM

Underwater Acoustic Communication

12/10/13 Rochester Institute of Technology 46

MSD 2 Schedule