White Paper. White Paper. *** A Critical Look at FreeDV plus Video *** October 2013

Transcription

1 White Paper *** A Critical Look at FreeDV plus Video *** by Rick Peterson, WA6NUT October 2013 =========== THE PROBLEM =========== First-time users of FreeDV plus Video (or even would-be users) might wonder why the video quality isn't better. The obvious answer is that the available bandwidth limits the image resolution to just 36 lines. The problem isn't with the ZL2AFP video software -- it's because of the interaction between the video and FreeDV sub-carriers. As a result, adding digital voice (DV) makes the video image more noisy, and adding video reduces the SNR of the digital voice component. AN ALL-DIGITAL SOLUTION? How does this happen? It would be possible to improve the video image and digital voice SNR by completely redesigning the video subsystem, replacing the present analog signal format with a digital format. The redesigned video subsystem would put a DQPSK (differential QPSK) stream on each subcarrier, with one luminance subcarrier for each line in the image (36 total). Chrominance data would be transmitted on 13 DQPSK subcarriers. The symbol rate for each subcarrier would be 50 symbols/sec, with 2 bits per symbol. The video software and digital voice software would be integrated into one package, so that the DV and video signals would share common soundcard software buffers (providing phase coherency, assuring complete orthogonality between the DV and video subcarriers). But more work is required to determine whether such a system would be feasible. The present system is a less than ideal pairing of analog (video) and digital QPSK (digital voice) subsystems. The literature refers to this as a "frequency division multiple-access (FDMA)" scheme. = THE ORTHOGONALITY PROBLEM = An earlier version of the video subsystem used Hz subcarrier spacing, and video and DV quality were barely acceptable (see screenshot #1). To improve the orthogonality with respect to the DV subcarriers (75 Hz spacing), the video subcarrier spacing was reduced to 37.5 Hz (see screenshot #2). The result was improved DV SNR and video image quality. Improvement was also seen with less scatter in the constellation diagram. But the improvement in the DV SNR and image quality were not as great as one would hope for. One limiting factor is that the video subcarriers are FM'ed. The luminance subcarriers can shift as much as 4 Hz from an all-black image to normal full-color images. So the effort made to provide "orthogonal" spacing of the video subcarriers with respect to the DV subcarriers is defeated by FM'ing of the subcarriers with peak deviation exceeding 10% of the subcarrier spacing! Page 1

2 But it is important to note that non-orthogonal frequency division multiple-access (FDMA) schemes have been successfully used (in satellite and cellular systems, for example). ======== IM PRODUCTS -- THE REAL PROBLEM? ======== The presentation by Dr. Leila Ribeiro suggests that a primary concern of FDMA systems is IM products. The Appendix shows how FreeDV plus Video can be modified to remove 3rd order IM products, removing interaction between the FreeDV subcarriers, the FFT-TV chroma subcarriers, and the FFT-TV luminance subcarriers). ========== CONCLUSION ========== The last conclusion in the paper by Wang, Xiao and Ping states: "Hybrid schemes can combine the advantages of both orthogonal and nonorthogonal techniques and provide promising solutions for future wireless cellular communications." It seems that FreeDV plus Video might fit this description! So the present FreeDV plus Video system is the result of combining two dissimilar modes, requiring only minor modification of the video component. Instead of asking, "Why the noisy video?" and, "Why the degraded DV SNR?", one might exclaim, "It's a wonder that it works as well as it does!" Interested experimenters are invited to become involved in the FreeDV plus Video project -- there's lots of opportunity to improve the software and the hardware mechanization. ======== APPENDIX ======== Removing 3rd Order IM Products from FreeDV plus Video The presentation by Dr. Leila Ribeiro suggests that 3rd order IM products can be eliminated by careful placement of user frequencies. For the FreeDV plus Video signal, there are only two users, FreeDV and ZL2AFP's FFT-TV signal (we make the chrominance and luminance one contiguous group -- as originally designed by ZL2AFP). The problem is to set the width of the guard band between f2 and f3 such that the IM products do not overlap the FreeDV or FFT-TV subcarriers. See Figure 1. FreeDV (1400) FFT-TV _ _ _ _ _ _ _ _ _ _ _ f f f f Figure 1: The FreeDV and FFT-TV signals Page 2

3 For the FreeDV signal: f1 = khz f2 = khz For the FFT-TV signal: f4 - f3 = 1.8 khz We can set f7 = 2f3 - f2 = f (sets the space between f4 and f7). We then calculate the guard band f3 - f2 = khz, giving f3 = khz (transmux B104) and f4 = khz (transmux B152). Using the method shown in the Ribeiro presentation, we calculate: See Figure 2. f5 = 2f1 - f4 = khz f6 = 2f2 - f3 = khz f7 = 2f3 - f2 = khz f8 = 2f4 - f1 = khz IM IM Products FreeDV (1400) FFT-TV Products _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ f f f f f f f f Figure 2: Signals with IM Products Added Note that f7 - f4 = 75 Hz, and that the baseband frequency corresponding to f4 is 5.7 khz (= the audio and RF bandwidth required). The width of the guard band (= f3 - f2) is khz. CONCLUSION The problem with 3rd order IM products is solved by increasing the signal bandwidth from 3.5 khz to 5.7 khz. Some HF transceivers provide TX and RX bandwidth of 3.5 khz or more, but very few provide 5.7 khz bandwidth (FlexRadio is one example). So the trade-off is: 1. We can have better voice and video quality, but only with a limited selection of HF transceivers with sufficient bandwidth (5.7 khz), or 2. We can have lower-snr voice and noisier video, but with a larger selection of HF transceivers with sufficient bandwidth (3.5 khz). Here we have an example of "real world" constraints getting in the way of performance! Of course, the present FreeDV plus Video system is based on the second alternative. An additional advantage is that the narrower bandwidth is less susceptable to QRM and creates less QRM on the crowded HF bands, and easily meets the FCC bandwidth requirements for a Page 3

4 "multiplexed image and phone emission." BIBLIOGRAPHY Rowe, David, "FDMDV Modem" Dr. David Rowe, VK5DGR, has a description of the FreeDV signal ("1400" version) at: Wikipedia, "Quadrature Amplitude Modulation" The Wikipedia entry for QAM (Quadrature Amplitude Modulation) is helpful (FreeDV is a QAM system): LitePoint Corp., "Practical Manufacturing Testing of OFDM Wireless Devices" This document, pp , explains the QAM constellation (scatter) diagram: %20Devices_WhitePaper.pdf Ribeiro, L., "Satellite Communications -- Multiple Access" This PowerPoint presentation (for a course at George Mason University) gives an overview of FDMA with a brief discussion of IM products: These papers discuss multicarrier FM systems (ZL2AFP's "FFT-TV" is a multicarrier FM system): Wachs, Marvin, "Analysis of Adjacent Channel Interference in a Multicarrier FM Communications System" (pp ) Chitre, N. K. M., and Fuenzalida, J. C., "Baseband Distortion Caused by Intermodulation in Multicarrier FM Systems" (pp.76-89) Wang, P., Xiao, J., and Ping, L., "Comparison of Orthogonal and Non- Orthogonal Approaches to Future Wireless Cellular Systems" This paper compares orthogonal and non-orthogonal systems (FreeDV plus Video is a non-orthogonal FDMA system, actually a hybrid formed from an orthogonal component -- FreeDV, and a non-orthogonal component -- ZL2AFP's FFT-TV): comparison_wang06.pdf LAST REVISED Page 4

5 This document last revised 10/4/13 Page 5

6

7