Moorea 2007 UCSB UWA Modem Results

Transcription

1 Moorea 2007 UCSB UWA Modem Results Presented by Tricia Fu

2 Field Test Staff Faculty: Prof. Ryan Kastner GSR s: Bridget Benson (Captain), Daniel Doonan (hardware), Tricia Fu (software), Chris Utley (mariner). Additional Captains: Yvette Ralph, Dr. Andrew Brooks Affiliated Faculty: Prof. Hua Lee, Prof. Ron Iltis Affiliated GSR: Taehyuk Kang (OFDM)

3 Outline Overview of different test types (Direct Sequence Spread Spectrum Matching Pursuits (DSSS MP) modem, raw data collection test, received vector (RV) collection test). Discussion of 4 main sets of test results: Unanchored transducer testing: Cook s Bay + Viapahu Lagoon Anchored transducer testing: Viapahu Lagoon - Tests A, B, C Preliminary underwater OFDM test Results summary Future work

4 Test Outline DSSS MP Modem Test Raw Data Collection Test (for OFDM, other offline processing) RV Data Collection Test (for in-band SNR, other offline processing)

5 DSSS MP Modem Test Modem tests ran over message symbols (240 packets) 18 training symbols transmitted between message packets 8 training symbols used at a time for symbol synchronization 43 message symbols per packet

6 MP Modem Test Details F adc F chip F center M-Sequence Length Gold Code Length Symbol Duration Time guard interval F sym F bit 96 khz 6 khz 24 khz 7 chips 8 binary symbols => 56 chip sequences ms ms baud ~ 54 symbols/sec bps ~ 161 bits/sec

7 Raw Data Collection Test Collect unprocessed ADC samples on the C6713 DSP board using same hardware used for modem testing. Data sampled at 96 khz. Length 2e6 element data arrays per test, either with Tx on or Tx off. Used primarily for initial OFDM testing.

8 RV Data Collection Test Collect processed received vector (RV) data on the C6713 DSP board for future offline processing. Generally 3.416e6 complex samples ~ 1e4 symbols worth of data gathered, about same duration as a modem test. Used for calculating in-band input SNR and future offline processing.

9 Unanchored Transducer Test Results Locations: Cook s Bay (ranges: m) Viapahu Lagoon (ranges: 28 m) Boats not anchored for Cook s Bay test. Transducers suspended from rope tied to boat in both locations.

10 Unanchored Transducers: Cook s s Bay Setup Transducers hanging from side of boat. Water ~ 30 m deep (boats not anchored). High estimated Doppler spread mostly due to transducer movement, and boat movement.

11 Unanchored Transducers: Cook s s Bay Test

12 Unanchored Transducers: Viapahu Lagoon Setup Right after the Cook s Bay testing, modem testing moved to shallower Viapahu Lagoon for preliminary trials. Hardware setup essentially the same, except boats now anchorable. 2 successful tests in this new environment are shown in the same results table as the Cook s Bay results.

13 Unanchored Transducers: Viapahu Lagoon Test

14 Unanchored Transducers Test Results Table MP Modem Test Test # Dist (m) NF SER DS med. filt (Hz) NF range Mean NF 1 L1: 25 Est. NF L1: 25 Est. NF L1: 50 Est. NF L1: 50 Est. NF L1: 60 Est. NF L2: 28 Est. NF 9.7x L2: 28 Est. NF Locations: L1: Cook's Bay, 30 m floor depth L2: Viapahu Lagoon, 2.3 m floor depth Note: High Doppler spread, high SNR, low # of paths estimated by the modem. RV Test SNR (linear) SNR (db)

15 Explanation of Test Results Table Test # Dist (m) NF DS med. filt (Hz) Test label a number indicating a test s order in a sequence of tests; may contain references to additional notes. Contains a label and/or number, indicating distance between transducers, and possibly a description of the transducers surroundings. Path cancellation type: estimated number of paths (NF), or fixed SER Symbol Error Rate (0 SER 1) NF range Doppler Spread (DS, Hz), calculated by applying a five element median filter to the Doppler spectrum averaged over all delays/arrival times. Range of actual number of multipaths cancelled by the modem. Mean NF Average number of paths guessed by the modem over 240 packets (10320 symbols).

16 Unanchored Transducer Movement Transducer video Location: Viapahu Lagoon, Moorea Anchored boat Unanchored modem

17 Unanchored Transducers - Viapahu Lagoon Channel Estimate Plots Moorea - Viapahu Lagoon Unanchored Transducers - Test 6 contour plot, estimated ambiguity function x White regions indicate gain 1 below 3 colorscale minimum Frequency f (Hz) Channel delay L (ms) Ambiguity Function Averaged Multipath Intensity Profile (MIP) Multipath Intensity MIP, Moorea - Viapahu Lagoon Unanchored Transducers - Test 6, Est. NF, Nf = 1 to 3, SER: % Symbol Delay (in ms) Moorea - Viapahu Lagoon Unanchored Transducers - Test 6 multipath intensity profile averaged over all symbols 0.62 m 1.25 m R m R meters = delay L * 1.5 m/msec Range between transducers = 28 m Delay L (msec) Multipath Intensity Profile (MIP) 0

18 Anchored Testing Boats in shallower water in Viapahu Lagoon so they could be anchored. Came up with method to anchor the transducers as well. 3 series of tests: A (125 m, 280 m), B (330 m, 440 m), C (50 m, many coral bommies)

19 Anchored Testing - Viapahu Lagoon Test Set A: 125m

20 Viapahu Lagoon Test Set A Setup Shallow water, sandy bottom, few bottom obstructions. Water < 3.3 m deep at both Rx (receiver) and Tx (transmitter) transducers. Transducers tied to cement blocks, cement blocks sunk to lagoon floor. 2 ranges tested successfully: 125 m, 280 m. Raw OFDM data also captured for offline processing.

21 Viapahu Lagoon Test Set A: 280 m Rx Boat Tx Boat Tx Boat

22 Viapahu Lagoon Test Set A Results MP Modem Test Test # Dist (m) NF SER DS med. filt (Hz) NF range Mean NF 1* L1 Est. NF 7.2x L1 Est. NF 7.8x L1 Est. NF 2x L1 5 8x L x L L x L1 Est. NF 5.2x L2 Est. NF 1.0x L2 Est. NF 3.0x L x ** L x RV Test Dist (m) SNR (linear) SNR (db) * = Tx transducer fell over during test ** = problem with hardware Distances: L1: 125 m ± 8.3 m, L2: 280 m

23 Anchored Testing - Viapahu Lagoon Test Set B Test Site GPS Coordinates: 330 m: Rx: S ', W ' Tx: S ', W ' Range: 333.6±7.2 m 440 m: Rx: S ', W ' Tx: S ', W ' Range: 440±7 m

24 Viapahu Lagoon Test Set B Pics: : 330 m, 440 m Rx Boat Tx Boat Tx Boat

25 Viapahu Lagoon Test Set B Results MP Modem Test Test # Dist (m) NF SER DS med. filt (Hz) NF range Mean NF 1 L1 Est. NF 1.98x L1 Est. NF 3.1x L x L x L L L L L2 Est. NF 6x L2 Est. NF 6x L x RV Test Dist (m) SNR (linear) SNR (db) Distances: L1: 333 m +/- 7.5 m L2: 440 m +/- 25 m SER averaged < 1% at both 330 and 440 m. Few obstructions on lagoon floor. Doppler spread (after median filtering) generally on the order of 0.01 Hz. Only 2 watts power fed into transducers.

26 Viapahu Lagoon Test Set B Channel Estimate Plots Frequency f (Hz) Moorea - Viapahu Lagoon Series B - Test 5 contour plot, estimated ambiguity function White regions indicate gain below colorscale minimum Channel delay L (ms) Range between transducers = 330 m Multipath Intensity m 0.87 m 1.37 m x Symbol Moorea - Viapahu Lagoon Series B - Test 5 multipath intensity profile averaged over all symbols 1.87 m 2.37 m 2.75 m MIP, Moorea - Viapahu Lagoon Series B - Test 5, Fixed NF, Nf = 2 to 5, SER: 0.0% Delay (in ms) R meters = delay L * 1.5 m/msec Delay L (in ms) R m

27 Anchored Testing - Viapahu Lagoon Test Set C Setup Shallow water, in coral bommies area near MRB Range: 50 m When both transducers set on lagoon floor, no direct path. Successful tests with Tx transducer set on coral bommie, Rx transducer set on lagoon floor.

28 Viapahu Lagoon Test Set C Pics: : 50 m, Bommies

29 Viapahu Lagoon Test Set C Results MP Modem Test Test # Dist (m) NF SER DS med. filt (Hz) NF range Mean NF 1* L1 Est. NF 8.7x L1 Est. NF L1 Est. NF L L L L * = Tx and Rx transducer both set on lagoon floor, thus results imply no direct path. In all other tests, Tx placed on coral bommie, and Rx left on lagoon floor. Distance - L1: 50 m No SNR measurements available, but assumed to be high (> 15 db). Lagoon floor between Tx and Rx transducers littered with coral bommies. Doppler spread (after median filtering) again generally on the order of 0.01 Hz.

30 Viapahu Lagoon Test Set C Channel Estimate Plots Frequency f (Hz) Moorea - Viapahu Lagoon Series C - Test 5 contour plot, estimated ambiguity function White regions indicate gain below colorscale minimum Channel delay L (ms) x Symbol MIP, Moorea - Viapahu Lagoon Series C - Test 5, Fixed NF, Nf = 2 to 5, SER: 0.0% Delay (in ms) Range between transducers = 50 m

31 Preliminary Underwater OFDM Test Used same hardware as MPbased UWA Modem. Used raw data test to collect ADC samples for offline processing. # of carriers 512 # of used carriers # of data bits in one OFDM symbol Data rate Carrier frequency System bandwidth # of OFDM symbols in a packet bits kbps 24 khz 4 khz (22-26 khz) 25

32 Underwater OFDM Test Results Test # Dist (m) LDPC Coded BER BER Conventional Gain BER after Decoding MP No 1.00% 1.36% 26.20% N/A Yes 1.33% 1.94% 31.44% 0% No 0.22% 0.37% 40.54% N/A Yes 0.31% 0.55% 43.64% 0% Transmitted signal Received signal

33 Results summary Doppler spread (averaged, median-filtered): High(averaging > 1 Hz) when boats and transducers are unanchored. Generally averaging on the order of 0.01 Hz with anchored transducers in Viapahu Lagoon. Modem Performance: DSSS MP modem: average(ser) < 1% OFDM modem: 0% BER in 2 tests with LDPC

34 Future Work Repackaging the MP-based modem for actual deployment Higher data rates with underwater OFDM

35 Acknowledgements UC Santa Barbara Marine Science Institute: Prof. Sally Holbrook, Prof. Russ Schmitt, Dr. Andrew Brooks, Yvette Ralph UC Berkeley Gump Station Keck Foundation, International Foundation for Telemetering, US Navy, Sonatech, and the UC MICRO Program