High Data Rate, Reliable Wideband HF Communications Demonstration

Transcription

1 High Data Rate, Reliable Wideband HF Communications Demonstration Maureen P. Scheible 1, Dr. Lucien J. Teig 1, John D. Fite 1, Kevin M. Cuomo 1, Janet L. Werth 1, Glenn W. Meurer 1, Nathan C. Ferreira 1,2 Cecelia R. Franzini 1,2 1 The MITRE Corporation Bedford MA, McLean VA, and Rome NY, United States of America 2 Worcester Polytechnic Institute Worcester United States of America Abstract - Assured beyond line-of-sight (BLOS) communications is challenging yet essential for our warfighters. Military and civilian systems rely on a combination of high data rate satellite connectivity as well as low data rate High Frequency (HF) skywave communications. While satellite communication provides high data rate connectivity, there are vulnerabilities that may degrade or disrupt service. To ensure critical communications are maintained, we are investigating the capability of higher bandwidth and higher data rate HF communications applying polarization diversity Multiple-Input Multiple-Output (MIMO) concepts. In addition, we are developing a multi-mode wideband HF channel simulator. We have successfully demonstrated short-hop (< 75 km) one-way HF skywave communication by simultaneously transmitting independent messages on two orthogonal circular polarization channels (O- and X-modes) for increased capacity and reliability. The communication waveforms we developed include higher capacity and bandwidth alternatives to the wideband waveforms in MIL-STD-188/110C Appendix D. In this paper, we demonstrate the results of our over-the-air (OTA) experiments and compare our performance against theoretically-predicted capacity for each waveform. We also describe the wideband HF channel simulator. Keywords: Ionosphere, HF, MIMO, polarization diversity, communication waveforms I. INTRODUCTION In today s data-rich world, BLOS communications are predominantly accomplished by satellite owing to the very high data rate offered by such systems. While satellite communication is very capable, it is also expensive and can be vulnerable to degradation and disruption. In order to maintain connectivity in challenging or denied environments, we are investigating increasing the capacity, robustness, and reliability of the HF communications channel. HF communication has generally been limited to narrower bandwidths and lower data rates compared to satellite communication. Traditional HF challenges also include channel fading and the need for large antennas. There have been a number of efforts focused on increasing HF link capacity that include larger bandwidth, high-constellation QAM signals (12-24 khz) [1], OFDM signaling with up to 256-QAM and advanced error coding [2], multi-carrier 1x2 MIMO with time, frequency, and polarization diversity and OFDM signaling [3], and 2x2 HF MIMO with polarization diversity [4]. The above have achieved spectral efficiency ranging from 1 to 4 b/s/hz under favorable conditions and high power. Efforts have included field trials in both the United States and Europe. Commercial industry is also making advancements in higher data rate HF systems for tactical military applications using the MIL-STD-188/110C Appendix D wideband waveforms. A new waveform has been developed for 2- antenna MIMO through the use of spatial multiplexing. Link simulations on a 48-kHz channel that use channel measurements from an NVIS collection campaign show a doubling of data rate and improvement in bit error rate (BER) by up to 15 db SNR in comparison to single antenna communication [5]. Here we demonstrate wideband HF one-way communication using a short-hop (< 75 km) skywave link. We apply new techniques and technology to wideband HF (WBHF) communications (up to 96 khz). We develop WBHF waveforms that are not constrained to the MIL-STD-188/110C Appendix D family of wideband waveform standards. We apply polarization diversity MIMO technique coupled with advanced waveform coding for increased capacity and interference rejection. We leverage improved understanding and adaptation to HF channel phenomenology to emphasize low power operations with increased bandwidth, capacity, and reliability. II. SYSTEM DESCRIPTION A. HF Communication Waveform Design The goal of our HF communication waveform designs is to communicate reliably at a rate of 1 to 4 bits/s/hz. To achieve this capability we exploit polarization diversity to simultaneously transmit on both the X- and O-propagation modes of the ionosphere using uncorrelated waveforms that carry independent messages. This approach approximately doubles the processing bandwidth while improving robustness to multi-path and channel fading. In figure 1, we show a high level block diagram depicting our approach. Note that two independent data streams are transmitted by processing them in parallel at baseband on the two separate polarization channels (blue and red boxes). A combination of Reed-Solomon and convolutional forward error correction coding (FEC) is used to improve the BER performance in the presence of both burst-like and multi-path fading that typically occurs over HF communication channels.

2 The coded data streams are then mapped to baseband signal constellations for low data rate (OFDM), medium data rate (BPSK), or high data rate (QAM) communications. The resulting data packets are assembled into frames and orthogonal preamble waveforms are added to the beginning of each frame to distinguish the two polarization channels. The two baseband signal frames are then up-converted in a dualchannel Ettus N210 USRP software defined radio (SDR) and simultaneously transmitted using orthogonal circular polarizations (e.g., left-circular (LC) for channel 1 and rightcircular (RC) for channel 2). The circular polarizations are actually achieved by using two collocated mutually perpendicular horizontal dipoles excited in quadrature for each of the desired signals (see section III in this paper). We refer to the two dipoles as H and V in this section for simplicity and brevity only. The circular polarizations result from transmitting simultaneously on the V and H dipoles using the appropriate +/- 90 degree phase shifts between dipoles. The goal of our HF communication waveform designs is to communicate reliably at a rate of 1 to 4 bits/s/hz. To achieve this capability we exploit polarization diversity to simultaneously transmit on both the X- and O-propagation modes of the ionosphere using uncorrelated waveforms that carry independent messages. This approach approximately doubles the processing bandwidth while improving robustness to multi-path and channel fading. In figure 1, we show a high level block diagram depicting our approach. Note that two independent data streams are transmitted by processing them in parallel at baseband on the two separate polarization channels (blue and red boxes). A combination of Reed-Solomon and convolutional forward error correction coding (FEC) is used to improve the BER performance in the presence of both burst-like and multi-path fading that typically occurs over HF communication channels. The coded data streams are then mapped to baseband signal constellations for low data rate (OFDM), medium data rate (BPSK), or high data rate (QAM) communications. The resulting data packets are assembled into frames and orthogonal preamble waveforms are added to the beginning of each frame to distinguish the two polarization channels. The two baseband signal frames are then up-converted in a dualchannel Ettus N210 USRP software defined radio (SDR) and simultaneously transmitted using orthogonal circular polarizations (e.g., left-circular (LC) for channel 1 and rightcircular (RC) for channel 2). The circular polarizations are actually achieved by using two collocated mutually perpendicular horizontal dipoles excited in quadrature for each of the desired signals (see section III in this paper). We refer to the two dipoles as H and V in this section for simplicity and brevity only. The circular polarizations result from transmitting simultaneously on the V and H dipoles using the appropriate +/- 90 degree phase shifts between dipoles. Fig. 1. High level signal flow used in our design and transmission of high data rate, polarization diverse HF communication waveforms. Figure 2 shows the communication frame structure and tabulates the achievable data rates as a function of bandwidth and modulation type that we have successfully tested. The preamble is a broadband wide-instantaneous bandwidth (WIBW) waveform that is used for range, Doppler, and phase alignment of the data packets in the receiver. The preambles for channels 1 and 2 are orthogonal waveforms that enhance MIMO separability and are robust to multi-path and channel fading. The packet identification (PID) record contains metadata that describes the various waveform parameters, including the type of modulation employed for the data symbols that follow. Typical frame lengths vary from 0.5 to 5 seconds depending on channel conditions and other factors. In our experiment, the preamble, PID, and data symbol lengths were matched to the receiver s sampling frequency (195 khz) allowing efficient processing by using power-of-two FFTs. Figure 2(b) shows the achievable data rates as a function of bandwidth and modulation type assuming perfect dual channel separability using orthogonal circular polarizations. The sparse OFDM waveform uses a subset of the total number of available frequency slots for transmission of the data symbols; this results in lower data rates (< 1 bit/s/hz), but is still high enough for many voice applications. Because the information bandwidth is substantially less than the total communication bandwidth, direct sequence spread spectrum techniques can be employed to significantly reduce the power spectral density and improve the performance of OFDM in the presence of severe interference. For higher data rate applications the BPSK and 4-QAM waveforms achieve 2 and 4 bits/s/hz respectively, but are, of course, more susceptible to noise and other forms of interference. In particular, the 4-QAM data rate of 375 kb/s is suitable for many real-time video applications. Preamble (NFFT = 16384) (a) Communication frame structure 84ms PID (8192) Data S 1 (8192) Variable length... b/s/hz Bandwidth (khz) Sparse OFDM < (b) BPSK QAM MIMO Data Rate (kb/s) Low rate (voice) Medium rate High rate (video) Fig. 2. (a) Communication frame structure showing preamble, packet identification (PID), and data symbols. (b) Achievable data rates as a function of bandwidth and modulation type (using dual channel polarization MIMO).

3 B. Receiver Signal Processing Figure 3 shows a block diagram of the signal processing employed in the receiver. The Ettus N210 SDR simultaneously samples the signals at both the vertical and horizontal elements of the receiver dipole antenna. The signals are filtered and down-sampled to baseband with a sampling rate of 195 khz. Polarization matched filters separate the H and V channels into the O- and X-propagation modes. The two modes are processed in parallel to recover two independent data streams (figure 3 only shows channel 1). The baseband processing begins by reliably detecting the preamble or sync waveforms for each channel. This is accomplished using two-dimensional complex ambiguity function (CAF) processing to estimate the delay, Doppler, and phase of the channels as a function of time. This process is carried out for every received data frame so that the current channel conditions can be estimated and tracked over time. Typical sync waveform detections are shown in the lower left plot over a four second time period (note that the frame length in this case is about 1.5 seconds). The sync detections provide for accurate data packet alignment and phase compensation from which an N-point FFT follows to demodulate the received signal constellations (lower right plot). Note that the OFDM signal tones are clearly apparent in this example the tone positions and/or phases define the data symbols that were transmitted. While not explicitly shown in this figure, we have also developed a technique for detecting and nulling strong narrowband interferers. The nulling process is followed by spectrum despreading which produces a clean signal constellation for subsequent processing. The final step involves FEC decoding which attempts to correct for any bit errors that may have occurred during the received signal demodulation and bit recovery process. Rx dipole V augmented variations [6][7] combine the effects from different ionospheric layers and magneto-ionic modes in a single composite model. For the development of MIMO systems exploiting polarization diversity, we've developed a model that treats each component separately and accounts for the responses of polarized transmit and receive antenna arrays. The model for each dominant path has both a deterministic and statistical component. The deterministic component uses a simple ray-tracing model that assumes a time-varying floating virtual reflection point (RP). The floating RP behavior can be defined to simulate changing layer heights and tilt angles. The outputs of this model are a time-varying range (used to compute corresponding time-varying Doppler, delay, and propagation loss), direction of transmission (DOT), and direction of arrival (DOA) (see figure 4). The statistical component augments each dominant path by introducing Rician fading, Doppler spread, and delay spread due to channel dispersion and absorption [8]. The HF channel simulator models the transmit and receive antenna array responses as well as the separate RC and LC transmission over the two ionospheric modes, X and O. Antenna types modeled include cross-dipole and an electromagnetic vector sensor [9]. The MIMO algorithms are able to specify the transmit and receive weight to achieve optimum performance. Rx Data FEC Decoder (N-point FFT) coded bits Signal Constellation Demodulation 2D CAF processing of sync O Polar. matched X filters H V H Baseband Receive Processing 2D-CAF Processing of Sync Sync detections (3 frames processed) Rx d waveform Packet alignment (timing, Doppler, phase, Eq.) Signal Constellation Demodulation (OFDM case) Data packet coded alignment, bits + FFT processing Fig. 4. HF Channel Simulator processing flow Time-varying parameters derived from the floating reflection point model are used to modulate the channel response. The adaptive weight W controls the effective transmit polarization. III. OVER THE AIR EXPERIMENT Fig. 3. High level receiver signal processing flow illustrating the data recovery process. Polarization matched filters are used to recover the O- and X-mode transmission channels; 2D CAF processing aligns the data packets in range, Doppler, and phase, while Fourier and FEC decoder processing is used to subsequently recover the data bit streams.. C. HF Channel Simulation In support of the system design for the wideband HF MIMO solution and experiment, we have developed a new HF channel simulator (figure 4). Most HF channel models used for analysis and performance evaluation assume a single-input single-output (SISO) configuration. The Watterson model and A. System Design On 6 and 21 December 2013, we conducted outdoor MIMO HF waveform experiments between the MITRE campus in Bedford, MA and the Worcester Polytechnic Institute (WPI) campus in Worcester, MA. The MITRE Bedford campus served as the transmit site and WPI served as the receive site. The ground distance between the two sites is approximately 50 km. Each site was equipped with a cross-dipole antenna (drawing shown in figure 5) that supports two orthogonal polarizations via two inverted vee-dipoles (x- and y- polarized). The antenna elements are well matched over 4-10 MHz to 50 ohms and offset to avoid mutual coupling effects. The antenna exhibits an azimuthally omnidirectional antenna pattern for RC and LC polarizations with an antenna gain at zenith of > -7

4 dbi at 4 MHz and > 0 dbi above 7 MHz. The system is simple, low-cost and MITRE-fabricated, -assembled, and -setup. Fig. 5. Crossed Dipole antenna model and dimensions. Fig. 6. Over the air (OTA) experimental results. (a) Theoretical BER vs. SNR for 98 khz bandwidth waveforms; (b) OTA results for 49 khz bandwidth OFDM (9 kb/s) and (c) BPSK (94 kb/s) waveforms. For the OTA demonstration we transmitted over the skywave link simultaneously on the RC and LC polarization channels. The 2x2 MIMO system used only two Ettus N210 USRP radios, one on transmit and one on receive. The communication frame structure depicted in figure 2(a) was configured for each bandwidth and modulation option shown in figure 2(b). Each waveform configuration (total of nine) was stored in a MATLAB flat file and sent to the radio for non-realtime transmission for approximately 30 minutes. The experiment was conducted at 5 MHz between the hours of 3:00 5:30 PM EST. B. OTA Experimental Results As discussed above, the primary goal is to demonstrate high data rate reliable communications at rates of 1 to 4 bits/s/hz. Our OTA demonstrations aim to achieve this capability by exploiting polarization diversity to simultaneously communicate independent waveforms via the X- and O- propagation modes. The preliminary test results presented below prove the feasibility of this concept. In figure 6(a) we show theoretical BER vs. SNR curves for the 98 khz bandwidth OFDM, BPSK, and 4-QAM waveforms. The dashed lines show performance using a Reed-Solomon (2/3-rate) coder/decoder whereas the solid lines show the corresponding uncoded performance. This type of FEC lowers the SNR threshold by roughly 3-5 db as expected. As a first test of our multi-mode HF communications concept we operated the transmitters at low-power (~5W per channel) to evaluate the robustness of our waveforms and signal processing at low SNR levels. As an example, we show, in figures 6(b) and 6(c), a comparison of the theoretical performance to the OTA results for the 49 khz OFDM and BPSK waveforms. The blue and red data points show the estimated BER vs. SNR for both the O- and X-mode data streams, respectively. Note that the data points are close to the theoretical curves even at these low SNR levels. The BPSK waveform achieved a data rate of approximately 94 kb/s (2 bits/s/hz). These first test results are very encouraging and we plan to conduct multiple follow-on OTA experiments over a wider range of bandwidths, power levels, and distances between the transmit and receive sites. C. Example HF Simulator Output To validate the HF Channel Simulator, simulated results were compared to field measurements. The HF Channel Simulator parameters were selected to mimic the time-varying amplitude of the magneto-ionic modes and observed "cross-talk" resulting from non-ideal transmit and receive antenna responses. In the field experiment, two orthogonal signals were transmitted with RHC and LHC polarization respectively. As expected, the RHC and LHC transmissions travelled through two different ionospheric paths. The appropriate combining weights and correlation signals were used to separately measure each signal. Most likely due to imperfections in the transmit and receive RF chain, a component of each signal "leaked" into the other ionospheric path. As shown in figure 7, in the simulation, transmit weights and parameters of the Rician fading model were selected to mimic observed field test results. Also note the absence of fast fading in the measured field results, a benefit of separating the two transmission modes. Fig. 7. HF Channel Simulator Parameters were selected to mimic field test configuration and receive measurements. IV. CONCLUSION We conducted OTA field experiments that provided a proofof-concept for new high data rate and reliable HF communications technology. We successfully demonstrated short-hop one-way HF skywave communication by simultaneously transmitting independent messages on two

5 orthogonal circular polarization channels (O- and X-modes) for increased capacity and reliability. Dual polarization data was received at WPI for all waveform types. Post processing of the WPI data demonstrated communication waveforms at very low power were easily detected and processed. The low data rate waveforms (OFDM) prove the feasibility of reliable dual polarization HF communications. The medium and high data rate waveforms were also received but require proportionally more transmit power than low data rate waveforms as expected. Significant interference rejection capability was demonstrated with the low data rate waveform and the measured and theoretical BER versus SNR curves are in good agreement. An HF channel simulator was developed that incorporates a time varying channel model with both deterministic and statistical elements. The simulation was run against the measured data and showed good agreement with measured path performance and behavior. V. FUTURE WORK We intend to continue developing waveforms that are more robust to interference and noise and intend to test these over the air at the same bandwidths reported here and at even wider bandwidths. We are also developing non-traditional sounder waveforms in order to sound the channels and develop algorithms to switch frequencies automatically to get better link performance. At an appropriate stage, we will test twoway transmission and real time signal reconstruction. We also plan to develop smaller HF dual-polarized antennas for mobile application. VI. REFERENCES [1] Jorgenson, M. B., et al. Polarization diversity for HF data transmission. International Conference on HF Radio Systems and Techniques, Nottingham, UK July [2] Icart, I., et al. Design and demonstration of a very high data rate multimedia HF communication system. IRST MAY 2012, the 12 th International Conference on Ionospheric Radio Systems and Techniques. [3] Perez-Alvarez, I., et al. Experimental results on multicarrier MIMO HF Communciations. September 2011 IEEE. [4] Ndao, P. M., et al. Implementation of a MIMO solution for ionospheric HF (3-30 MHz) radio links., th European Conference on Antennas and Propagation (EuCAP) Pages [5] Daniels, R. C. and S. W. Peters. A new MIMO HF data link: designing for high data rates and backwards compatibility., Preprint for IEEE Military Communications Conference [6] Rec. ITU-R F.1487: Testing of HF modems with bandwidths of up to about 12 KHz using ionospheric channel simulators. (2000). [7] Furman, W. N., and E. Koski. "Standardization of an intermediate duration HF channel variation model." The Institute of Engineering and Technology 11 th International Conference on Ionsopheric Radio Systems and Techniques 2009: Volume 546 pages [8] Olivadese, D., et al. "A radar oriented ionospheric channel model based on ray-tracing theory." Radar Conference (EuRAD), 2010 European. IEEE, [9] Ko, C. C., J. Zhang, and A. Nehorai. "Separation and tracking of multiple broadband sources with one electromagnetic vector sensor." Aerospace and Electronic Systems, IEEE Transactions on 38.3 (2002): Approved for Public Release; Distribution Unlimited The MITRE Corporation. All Rights Reserved.