Designing the MIMO SDR-based LPD Transceiver for Long-range Robot Control Applications Grigoriy Fokin, Dmitry Volgushev, Artem Kireev, Danil Bulanov, Vladimir Lavrukhin Research and Education Center for Wireless Communications The Bonch-Bruevich Saint Petersburg University of Telecommunications Saint Petersburg, Russia lavrukhin@sut.ru Abstract There is a huge growth in robot control applications over the past years. One of the main challenges in such applications is the radio link between a control point and a robot. To improve the range of the link most available control systems use 433 MHz Low Power Devices (LPD) band for communication. Required distance between a robot and control point for our application is 10 km. Achieving this distance is a challenge when multipath propagation is present. Our Matlab simulation demonstrates that 2*2 Multiple Input Multiple Output (MIMO) Alamouti scheme allows to gain more than 10 db comparing with Single Input Single Output (SISO). Link budget calculations show that we can achieve even more than 10 km distance using 2*2 MIMO scheme. High precision Agilent hardware and SystemVue software were used for performance evaluation of the radio link and system architecture. To prototype the system we used USRP B210 Software Defined Radio (SDR) hardware due to huge variety of supportive platforms and convenient type of connectivity. Keywords SDR, 433 MHz, LPD, MIMO 2*2, Robot control I. INTRODUCTION This paper describes the current University research and development (R&D) project. The main goal of the project is to develop the 2x2 MIMO SDR-based 10 km distance transceiver in the 433 MHz ISM UHF frequency band. Current research activity analysis has shown the absence of available MIMO SDR-based long-range transceivers in 433 MHz ISM UHF band. That is why this field of R&D appears to be quite perspective. Most of existing 433 MHz band transceivers have a discrete component design. But we use the SDR solution to make the cognitive radio design possible in the future. A small section of the ISM 70-cm band near 433 MHz is available for license exempt wireless communication using low power devices (LPDs) [1], also called short-range devices (SRDs) [2]. We use Ettus Research B210 Universal Software Radio Peripheral (USRP) hardware to achieve 10 km range for low rate telemetry exchange in robot control applications. Such a telemetry system requires only several kbps data rate to control the robot movement and has to be zero tolerant to information errors and losses [4]. In this paper we show that by using dual-antenna transmit diversity techniques based on Alamouti Space-Time Block Coding (STBC) [3] telemetry exchange can be performed reliably. To assess the performance of the given system we used intensive simulations to evaluate BER for MIMO 2x2 Alamouti scheme that is intended to achieve 10 km range. Then we defined E b /N 0 threshold value and compute radio link budget. To verify the system performance we developed a testbed based on Agilent hardware and software. It offers a real time visualization of the signal travelling through the radio channel. This paper is organized as follows. In the Section II we present an analysis of commercially available SDR solutions and a short overview of Ettus Research B210 hardware capabilities. Section III describes the simulations to evaluate bit error probability for MIMO 2x2 Alamouti scheme. Section IV contains Radio Link Budget calculation. Section V describes the Agilent testbed used for performance evaluation of the radio link. Section VI shows USRP B210 board design and implementation of MIMO 2x2 Alamouti scheme. The last two sections present a future work and conclusion. II. SDR HARDWARE ANALYSIS There are only few commercially available MIMO SDR solutions in the world (Table 1). Most of the hardware has a 2*2 MIMO transceiver. Supported frequency range is wide enough for our aim. All the boards support 433 MHz LPD band and have a bandwidth of tens MHz. The given hardware has the similar technical characteristics. Therefore to prototype the system we have chosen the most popular and supportive Ettus Research USRP B210 platform. USRP B210 has two receive chains and two transmit chains [5]. It uses Analog Devices RFIC AD9361 single chip direct conversion transceiver as an integrated RF frontend. It is a cost-effective experimentation platform. B210 uses both signal chains of the AD9361 providing coherent MIMO capability, which makes it possible to develop MIMO 2x2 Alamouti scheme using the common USRP Hardware Driver (UHD) framework. UHD is an open source, cross platform driver that can run on Windows, Linux and Mac OS. It provides a common API 978-1-4799-5291-5/14/$31.00 2014 IEEE 978-1-4799-5291-5/14/$31.00 2014 IEEE 456
that is used by several software frameworks such as GNU Radio [6]. and then a modified version of these symbols in the second symbol period. TABLE I. Name (Vendor) Maveriq (Epiq Solutions) bladerf x40 (Nuand) bladerf x40 (Nuand) ASR-2300 (Loctronix) AD-FMCOMMS3- EBZ (Analog Devices) USRP B210 (Ettus Research) MIMO AVAILABLE MIMO SDR SOLUTIONS Frequency Range, MHz 2*2 100 6000 8 2*2 300 3800 n/a 2*2 300 3800 n/a Noise Figure, db FPGA Xilinx Spartan 6 LX150T 40KLE Altera Cyclone 4 E 115KLE Altera Cyclone 4 E 2*2 400 3800 n/a Xilinx Spartan 6 2*2 70 6000 <2,5 none 2*2 70 6000 <8 Xilinx Spartan 6 XC6SLX150 The onboard signal processing and control of AD9361 is performed by Spartan6 XC6SLX150 FPGA connected to a host PC using SuperSpeed USB 3.0 interface. III. MIMO 2X2 SIMULATION Multipath fading is a significant problem in radio communications. When the signal power drops dramatically, the channel is in a fade condition, which leads to high bit error rates (BER). MIMO technology significantly improves the link performance and reliability, especially in non-line-of-sight 433 MHz robot control application environment. Utilizing multiple element antennas (MEAs) both on transmit (Tx) and receive (Rx) sides of the communication link reveals exploitation possibility of both transmit and receive diversity schemes which yields a high quality of reception in terms of BER with a simple maximum likelihood (ML) decoding algorithm. Let us evaluate the performance of MIMO 2x2 Alamouti coding scheme with BPSK signal modulation in the case of slow Rayleigh fading channels when we have perfect knowledge of the channel at the receiver. The functional diagram of the system is shown in Fig. 1. This system consists of a transmitter (Tx) and a Receiver (Rx). Each of devices includes two antennas. There is a wireless channel between Tx and Rx. The relationship between the Tx and Rx is given by (1). y = H x + n (1) where y is a received signal vector, x is a transmitted signal vector, H is the complex channel matrix and n is a noise vector [12]. The Alamouti coding scheme involves a space-time block which has a pair of symbols sent in the first symbol period, Fig. 1. MIMO System Concept Two consecutive symbols x1 and x2 are encoded with the following space-time codeword matrix [3]: X = x 1 x 2 * x 2 x 1 * Under the assumption that the channel experienced by each transmit antenna is flat fading and independent from the channel experienced by other transmit antenna, each transmitted symbol in (1) gets multiplied by a randomly Rayleigh varying complex number h ij which is assumed to remain constant over two symbol periods: 1 1 y 1 h11 h12 n 1 1 1 h 2 21 h y 22 x1 n2 = + 2* * * 2* y h 1 12 h 11 x 2 n 1 2* * * 2* y h 2 22 h21 n2 where y 1 1 and y 1 2 are the received information at 1 st symbol period on 1 st and 2 nd receive antenna respectively; y 2 2 1 and y 2 are the received information at 2 nd symbol period on 1 st and 2 nd receive antenna respectively; h ij is the channel from i th receive antenna to j th transmit antenna; n 1 1 and n 1 2 are the noise at 1 st symbol period on 1 st and 2 nd 2 receive antenna respectively; n 1 and n 2 2 are the noise at 2 nd symbol period on 1 st and 2 nd receive antenna respectively. The diversity analysis is based on ML signal detection at the receiver side. Complex orthogonality of the Alamouti code (2) (3) 457
in (2) allows to simplify the ML receiver structure. Matlab simulation was executed [7]. Figure 2 shows the MIMO 2x2 Alamouti STBC performance in terms of BER. 2*2 MIMO Alamouti transceiver for a 10 km distance is feasible. Fig. 2. BER plot for 2 transmit 2 receive Alamouti STBC It is shown in figure 2 that the BER performance for 2 receive Alamouti STBC is much better than 1 receive case. For BER=0.0001 required (E b /N 0 ) req =10 db (line 14 in table II). IV. RADIO LINK BUDGET We calculated the radio link budget (table II). Main loss contribution in the radio link is path loss that was determined using Okumura-Hata Model [8] and plotted in Fig 3. We have chosen Ls = 150 db (line 4 in table II) as an operating point for rural area, because this is a primary area for our robot control application. Hardware gains include B210 transmitter power P t = -20 dbw (10 mw) (line 1 in table II), selected Tx antenna gain G t = 12 dbi [10], selected Rx antenna gain G r = 6 dbi [11]. Hardware losses include B210 receiver noise figure F = 8 db (line 8 in table II). Noise temperature [9] is computed according to T = (F-1) 290 = 1540 K (line 9 in Table II). Noise power spectral density is computed according to N 0 = k T = -197 dbw/hz (line 10 in Table II), where k is Boltzmann constant. Bitrate required for robot control application telemetry transmission is set to R = 1,2 kbps (line 12 in Table II). It is enough for transmitting a simple movement control commands, e.g. right, left, accelerate, stop. Operating (E b /N 0 ) oper value is defined from P r /N 0 value according to (E b /N 0 ) oper = P r /N 0 /R (line 13 in Table II). Communication system feasibility in terms of link margin (line 15 in Table II) considering (E b /N 0 ) req point on the BER curve is defined according to M = EIRP + G r (E b /N 0 ) req R N 0 L 0 (4) Fig. 3. Pathloss versus distance TABLE II. RADIO LINK BUDGET CALCULATION # Radio link budget parameter Denotation Value, db 1 Transmitter power (dbw) P t -20 2 Tx antenna gain (dbi) [10] G t 12 3 EIRP (dbw) EIRP -8 4 Pathloss (db) L S 150 5 RIP (dbw) RIP 158 6 Rx antenna gain (dbi) [11] G r 6 7 Received power (dbw) P r -152 8 Noise figure (db) F 8 9 Noise temperature () T = 1540 K 10 Noise spectral density (dbw/hz) N 0-197 11 Received P r/n 0 (db-hz) P r/n 0 45 12 Bitrate (db-bit/s) R = 1,2 kbps 31 13 Operating (E b/n 0) oper (db) (E b/n 0) oper 14 14 Required (E b/n 0) oper (db) (E b/n 0) req 10 15 Margin 4 V. AGILENT PERFORMANCE EVALUATION To verify the simulated 2*2 MIMO Alamouti system and evaluate the system architecture we use our laboratory facilities. The laboratory complex consists of two vector signal generators, two spectrum analyzers, channel emulator and SystemVue software. SystemVue is an electronic design automation environment for electronic system-level design. It works together with measurement equipment. It is possible to upload the simulated signals from software into the vector generator or channel emulator and to play signal on the given radio frequency. The channel emulator has several built-in radio effects: calibrated additive white Gaussian noise (AWGN), fading, multipath and Doppler. These options give us a unique opportunity to simulate the radio channel and move closer to the real radio channel in highly dynamic changing environment. Therefore algorithms and architecture of the designing device can be tested before first prototype is made. It helps to find bugs in algorithms and optimized parameters on early stages of design process. Resulting radio link budget margin is positive (line 15 in table II) which means that designed 433 MHz LPD 458
Fig. 4. Test-bed functional diagram; MXG vector generator, EXA spectrum analyzer, PXB channel emulator, VSA software for vector analyze, VSA files recorded signal The general conception of the performance evaluation of the radio link and system architecture, consist of three steps. Fig. 4 shows realization of this conception. The first step is modeling the system architecture of the future device in SystemVue software. It provides the simulated signal, which will be very similar to the signal in the real physical device. It is possible to check the performance of algorithms and plot the following characteristic: constellation diagram, BER, eye diagram, impulse response, spectrogram. The second step include a connection of the SystemVue to the measurement equipment, uploading simulated signal into the generator, and playing it on the radio frequency through a cable or an antenna. The third step is uploading the simulated files into channel emulator, adding the channel effects: noise, Doppler shift and multipath. All effects are added to the baseband layer. At the second and third steps the spectrum analyzer with a vector analysis and demodulation features makes the output signal control. Spectrum analyzer records signals to files (VSA-type). These files can be then moved to the SystemVue software for digital signal processing and demodulation, for example, for counting bit error rate. VI. USRP B210 BOARD IMPLEMENTATION Section A below presents USRP B210 hardware development framework for MIMO 2x2 Alamouti scheme. Section B contains results of the trial radio link test we have obtained after Matlab and Agilent test-bed simulations with BPSK/QPSK demodulation using UHD. A. B210 Overview Block diagram of the USRP B210 board consists of three main parts (Fig 5): AD9361 Transceiver; Xilinx Spartan 6 FPGA; Cypress FX3 USB3 Controller. First part is based on AD9361 transceiver from Analog Devices. This block is responsible for RF front-end processing using two receive and two transmit channel. Processing combines RF-to-zero-IF conversion and backward (down/up conversion), gain, filtration, ADC/DAC and decimation/interpolation. The AD9361 is connected to FPGA via Digital Data Interface which operates in CMOS or LVDS mode. For data transfer between AD9361 and FPGA I\Q format is used. Second part is based on Xilinx Spartan 6 FPGA and represents the digital signal processing blocks. List of blocks is determined by the developer and depends on the specific problem. Since RF part combines 2RX and 2TX channels it is possible to create a 2x2 MIMO system. To achieve this developer must realize MIMO encoder/decoder. To work with wide range of standards (such as FM and TV broadcast, cellular, Wi-Fi, ISM, and more), developer have to implement a specific modulator/demodulator and encoding/decoding scheme. 459
Fig. 5. Block diagram of the USRP B210 board The USRP B210 uses USB3.0 Interface under Cypress FX3 control for PC connectivity. This controller has a 32-bit RISC ARM926 processor. Serial Peripheral Interface (SPI) is used as a serial control interface to configure all of the AD9361 digital and analog blocks. The ARM processor executes FX3 firmware code, which processes instructions from/to PC and translates them into SPI commands to/from AD9361 SPI control registers. The SPI lines pass through the FPGA, where they are level translated, and then head to the AD9361. At the early stage of the project to explore the possibilities of the B210 application, all of the modulation\demodulation blocks were implemented in software using GNU Radio [6]. B. B210 hardware experiment It was essential to spend a few test experiments to create SDR applications development methodology based on B210 and GNU Radio. Choosing between Linux and Windows we decided to use Linux as the host operating system due to full support for Ettus Research USRP devices. We had to check several Linux distributions before one was found that fitted perfectly with B210. Ubuntu 12.04 was chosen because of stable work with combination of GNU Radio and UHD (B210 board). First, we realized transmitter using standard blocks of GNU radio library that can generate signals with different types of modulation (BPSK, QPSK and FSK). Spectrum analyzer by Agilent technologies was applied to check transmission results in the form of high-resolution constellation diagrams and spectrums. Since, GNU Radio supports UHD, we could control several parameters of zero intermediate frequency tract, such as operational frequency, sample rate, gain, low-pass filter corner frequency, decimation factor and more. To get correct results on the screen of spectrum analyzer, we faced some problems with adapting parameters between GNU Radio and Agilent. For example, scheme of transmitter in GNU radio consists symbol rate parameter and Agilent software has sample rate parameter and matching of these dimensions caused difficulties in getting constellation diagrams. All in all, hardware experiment consisted of PC with GNU radio transmitter block scheme connected with B210 board by USB cable and Agilent Spectrum analyzer with proper aerial for receiving signal. As a result we have got spectrums (Fig. 6, 7) of BPSK, QPSK and FSK modulations. VII. FUTURE WORK In the future work we will program the Xilinx Spartan 6 FPGA to make the MIMO 2x2 Alamouti modulator/demodulator and encoding/decoding scheme. After that we will continue laboratory tests to choose the best modulation and coding scheme. Then we will make a field trials in a rural area. 460
In the paper we proposed high precision Agilent hardware and SystemVue software test bench for performance evaluation of 2*2 MIMO radio link. We presented USRP B210 hardware development framework for our MIMO 2x2 Alamouti scheme and conducted trial radio link test with BPSK/QPSK demodulation. We used UHD capabilities in conjunction with GNU Radio software on the USRP B210 hardware. We verified the system on the proposed Agilent test bench. References Fig. 6. BPSK Constellation Diagram and Spectrum Fig. 7. QPSK Constellation Diagram and Spectrum VIII. CONCLUSION The aim of this paper is to show the design process of 433 MHz LPD band transceiver for long-range robot control applications up to 10 km distance. We suggested 2*2 Alamouti scheme for transceiver. We also computed the radio link budget for given distance. The results show that the communication system is feasible meeting all the requirements of the LPD band (especially low transmit power). [1] ETSI EN 300 220-1 V2.4.1 (2012-01), "Electromagnetic compatibility and Radio spectrum Matters (ERM); Radio equipment to be used in the 25 MHz to 1 000 MHz", ETSI, 2012. [2] ERC Recommendation 70-03 relating to the use of Short Range Device (SRD), European Conference of Postal and Telecommunications Administrations, CEPT/ERC 70-03, Troms, Norway, rev. 2014. [3] Alamouti, Siavash. "A simple transmit diversity technique for wireless communications." Selected Areas in Communications, IEEE Journal on 16.8 (1998): 1451-1458. [4] QoS Performance requirements for UMTS. Nortel Networks (2001). http://www.3gpp.org/ftp/tsg_sa/wg1_serv/tsgs1_03- HCourt/Docs/Docs/s1-99362.pdf [5] Ettus Research, ettus.com. [Online]. Available: http://home.ettus.com/ [6] GNU Radio, gnuradio.org. [Online]. Available: http://gnuradio.org/redmine/projects/gnuradio/wiki [7] Alamouti STBC with 2 receive antenna, dsplog.com. [Online]. Available: http://www.dsplog.com/2009/03/15/alamouti-stbc-2-receiveantenna/ [8] Recommendation ITU-R P.529-3 (1999-10), "Prediction methods for the terrestrial land mobile service in the vhf and uhf bands",itu-r,1999 [9] Sklar, Bernard. Digital communications. Vol. 2. NJ: Prentice Hall, 2001. [10] Polaris 433-7, radio-modem.ru. [Online]. Available: http://radio-modem.ru/antenna/polaris/polaris_433-7.htm [11] Polaris 433, radio-modem.ru. [Online]. Available: http://radio-modem.ru/antenna/polaris/polaris_433.htm [Accessed: Jul. 31, 2014] [12] Bialkowski, K. S., et al. "2x2 MIMO Testbed for Dual 2.4 GHz/5GHz Band."Electromagnetics in Advanced Applications, 2007. ICEAA 2007. International Conference on. IEEE, 2007. 461