2010 Second International conference on Computing, Communication and Networking Technologies
Implementation of a Low Cost Synthetic Aperture Radar using Software Defined Radio Deepthi Maheswari Chinnam, Madhusudhan J., Nandhini C., Prathyusha S.N., Sowmiya Sw., Ramanathan R*, Dr. Soman K.P. Amrita Vishwa Vidyapeetham, Coimbatore, India *
[email protected]
Abstract— GNU radio is a free open-source software toolkit for Radio. The spill over effect to the civilian sector is the building software radios, in which software defines establishment the of low cost community radios and the evolution transmitted waveforms and demodulates the received waveforms. of cognitive medium access (CMA) as a protocol for their In this paper an attempt has been made to explore the means to coexistence within independently evolving WLAN bands. Now use a Software Defined Radio (SDR) to implement a basic radar various SDR architectures are used for purposes ranging from system and then synthetic aperture radar. An experiment where distress signalling in natural calamities to enabling of in readings at two different scenarios (free environment and metal roaming by service providers. object) are taken into account and their plots are also given.international This has been attempted keeping in mind the exponential increase in II. SYNTHETIC APERTURE RADAR (SAR) chip computing power and the ability to upgrade a radio transceiver via software updates with a marginal investment, the RADAR (radio detection and ranging) is a widely used two features which makes such a foray attractive, technologysystem wise which uses electromagnetic waves for the detection of and cost wise. This attempt also takes us a step closer to objects, for terrain mapping and for weather forecasts. establishing the concept of a Cognitive radar which is software Synthetic aperture radar (SAR) is a type of imaging radar that signal processing intensive.
has wide range applications in obtaining the digital elevation models of the earth’s surface and in remote sensing [2]. The
Keywords— GNU Radio, Synthetic Aperture Radar, Software main advantage of SAR is the high resolution in spite of a defined radio (SDR) small antenna that is obtained by using the property of Doppler
shift of the backscattered signals. Another advantage of SAR I. INTRODUCTION AND BACKGROUND The flexibility of software based systems with regards to various use cases and adaptability to a variable environment will make them mainstream for many different applications. SDR reflects the convergence of two very dynamic technologies: digital signal processing and real-time downloadable software run on very fast microprocessors. GNU Radio is an open source software-defined radio project, and the Universal Software Radio Peripheral (USRP) is hardware designed specifically for use with GNU Radio. Together, these two technologies have been used to implement very sophisticated, low cost, SDRs. Since SDR and softwaredefined radar are really one in the same technologies, it stands to reason that GNU Radio and the USRP could be utilized to form a low cost radar sensor. In this paper, the viability of a prototype software-defined Synthetic Aperture Radar, built using the open source GNU Radio[3] and open specification USRP [4] has been discussed. first use SDR wasthe by Joint the USTactical Department of Defence andThe NATO whoof created Radio System (JTRS) initiative to develop a family of SDR and Cognitive
978-1-4244-6589-7/10/$26.00 ©2010 IEEE
over other imaging systems such as LIDAR (Light detection and Ranging) and optical imaging is that SAR can see through clouds. In SAR, both the amplitude and phase of the return signals are used for obtaining the images. The main blocks of an SAR system are the RF (radio frequency) front-end and the signal processor. The RF frontend consists of a transmitter, receiver, mixers, low noise amplifiers, analog-to-digital convertor (ADC) and digital-toanalog convertor (DAC). The analog signals from the receiver are digitized and are given as the input to the digital signal processor. The processor performs various frequency domain operations such as Fast Fourier Transforms (FFT) on the raw data and produces the final SAR image. Figure 1 shows the block diagram of a typical SAR.
Further, multiple software modules implementing different standards can be present in the radio system. The system can take up different personalities depending on the software module being used. Also, the software modules that implement new services or features can be downloaded over-the-air onto the handsets. The actual software being used is GNU Radio, which is an open source software-defined radio project. Fig.1: Block Diagram of a Typical SAR
The SAR images are two-dimensional in nature as both azimuth and range measurements are considered for image formation. The raw SAR data are points which are spread out in both range and along-track (azimuth) directions. These points are converted into a 2-D image by complex signal processing steps. Focusing is the process of correlation which produces a high resolution image. After focusing, multilooking is done so as to improve the resolution of the image. Due to the coherent nature of SAR imaging, speckle (grainy appearance) is produced due to the variance in the backscatter of different pixel cells. Speckle filtering is one of the most important signal processing steps. Geocoding, radiometric calibration and normalization are done so as to obtain a more accurate image of the terrain. III. NEED FOR THE PROPOSAL The digital systems are increasingly using programmable hardware modules at different functional levels. SDR technology aims to take advantage of these programmable hardware modules to build open-architecture based radio system software. SDR technology facilitates implementation of some of the functional modules in a radio system such as modulation or demodulation, signal generation, coding and link-layer protocols in software. This helps in building reconfigurable software radio systems where dynamic selection of parameters for each of the above-mentioned functional modules is possible. A complete hardware based radio system has limited utility since parameters for each of the functional modules are fixed. Implementation of functions such as FIR filter, sharp roll off and liner phase in hardware is not very easy, whereas these functions can be implemented in SDR as custom blocks and can be used wherever necessary. When we implement signal processing blocks in the hardware of conventional Radars, tuning and tweaking to a particular frequency may not be that accurate compared to the values that we can give in SDR. When an application requires the interoperability between different Radar systems rather than conventional radars, radars using SDR has better potential for interoperability. Moreover Radars using SDR have high sampling rate, high performance data converters (ADC’s & DAC’s), Wideband Tunable Analog RF Transceiver Front-Ends, Flash Memory and Wideband Linear RF Power Amplifiers. SDR technology promises to solve the problem of incompatible technologies and constant evolution of link layer protocol standards by implementing the radio functionality as software modules running on a generic hardware platform.
IV.S OFTWARE DEFINED RADIO A software defined radio (SDR) is a radio communication system where components (e.g. filters, mixers, amplifiers, modulators/demodulators, detectors) that have been implemented in hardware are instead implemented in software on a personal computer or other embedded computing devices (Fig.2). SDR technologies are attractive for communication systems because of reconfigurable and multimode operation capabilities. The reconfigurable feature is useful for enhancing functions of equipment without replacing hardware. Multimode operation is essential for future wireless terminals because a number of wireless communication standards will co-exist. Compared to traditional hardware-oriented approaches such as DSP and FPGA-based solutions, a true SDR base station is highly modular and enables a high degree of software portability and reuse, minimizing the amount of code to be rewritten to keep pace with advanced technologies. So, at a certain stage in the receiving chain the signal is digitized and passed to the software domain.SDR is divided into two subsystems-the hardware subsystem and the software subsystem. In general SDR consists of an antenna, an ADC and a software defined subsystem. Realizing such a device would require that each of the three following conditions is met: • •
•
The antenna should be capable to operate at the frequency of all radio signals of interest; The ADC and the DAC should have a sampling rate greater than two times the frequency of the signals of interest; The software subsystem should have enough processing power to handle the signal processing of all radio signals of interest.
In a typical SDR the hardware-defined subsystem consists of a wideband Radio FE that takes a portion of spectrum and shifts it to the IF prior to digitization. The software-defined subsystem receives the digital wideband signal, sends it to a set of digital down converters to isolate the required carrier, and proceeds with the demodulation.
Antenna (Daughte r Boards)
USRP
FPGA (ADC, DAC, USB Control, Digital Down/Up Conversion
PC/Laptop
Fig.2 Block Diagram of an Ideal SDR
V. SAR USING SDR The focus of this paper is not the design of a new radar sensor, but an investigation of the utility of existing hardware and software for software-defined radar applications. Therefore, the performance of the sensor hardware is evaluated in order to determine the applications for which the sensor is suitable.
A radar sensor must be capable of transmitting and receiving data such that the time between pulse transmission and reception can be known exactly. That is there should be time coherence and time-synchronization. A stream of digital data samples is time-coherent if a time value can be assigned to each sample such that the difference in the time values assigned to any two samples is equal to the difference between the actual times at which the samples are converted from, an analog signal. Therefore if the system is time-coherent, then the discrete data signal accurately represents its analog counterpart in time [1]. Thus the definition of time-coherence is concerned only with the duration between samples, and not the actual times at which these samples were converted. Two streams of digital data are said to lack time-synchronization if each stream is time-coherent within itself, but the two-streams are not time-coherent with respect to one another. In radar systems, time synchronization must exist between the transmit and receive data streams. Therefore, obtaining timesynchronization must be addressed explicitly.
Fig.3: Block diagram of radar software functionality
The radar transmit waveform is read in from a data file by the transmit signal processing block (TX block), which pushes this data into a software first-in-first-out (FIFO) buffer controlled by the GR framework. The GR framework facilitates the transfer of this FIFO data over the USB to the USRP where it is pushed into a FIFO on the FPGA. The USRP transmit signal processing path then pulls the data from this buffer as needed. An identical situation exists for the received data path except that the flow of data occurs in the opposite direction. In order to prevent underruns or overflows of information in the buffers, the Tx and Rx signal processing blocks must produce or consume the amount of data requested by the GR framework each time the block is called. Therefore, the transmitter and receiver must been enabled at all times.
or receive mode at a given time. However, this is not possible due to the asynchronous nature of the underlying systems.
Because the each system transmitting and one-half receiving simultaneously, signalispath can be allocated of the USB bandwidth. Note that one might be tempted to implement a scheme in which the system is in either transmit
the USB2 or port. the USRPfrequency) converts the RF signal the baseband IFThus, (intermediate signal. The to GNU Radio software is then used to manipulate the digital data.
VI.I MPLEMENTATION The Universal Software Radio Peripheral (USRP) is a device which turns general purpose computers into flexible SDR platforms [4]. The USRP consists of a motherboard with four high-speed ADCs and DACs and an Altera Cyclone EP1C12 FPGA. The ADCs/DACs are connected to the radio FEs (called daughter boards), while the FPGA is connected to a USB2 interface chip to a general purpose computer. The high speed general purpose processing, like down and up conversion, decimation, and interpolation are performed in the FPGA and the resulting digital signal is sent to a computer via
the implementation of SDRs [8]. The development of the software defined radar is done by representing the hardware radar by a graph where the vertices represent the various blocks and the edges represent the data flow between the blocks.
Fig.4: Internal and external view of USRP kit
By using the software defined radio (SDR), most of the hardware blocks in radar such as the modulator, waveform generator, demodulator and the signal processor in software in terms of their digital equivalents can be implemented. The equivalent digital filters of the analog filters used in the normal radar are found using the bilinear transformation technique. The GNU Radio software, an open source freeware, is used for
A. Transmit signal processing block A GNU Radio signal processing block accepts arguments which specify the file that contains exactly one pulse-repetition interval (PRI) of the radar waveform, the number of times the data in this file should be transmitted and the number of samples that the transmitter should wait before transmitting anything [1],[6]. Upon instantiation, the transmitter checks its state and responds accordingly. The system simply counts samples and transmits nothing. Once the requisite number of samples has been passed, it transmits the synchronization signal after which the system transmits a known signal.
Fig.5: Block diagram of software-defined radar in time synchronization mode
B. The Receive Signal Processing Block The constructor of this block accepts arguments which specify a pointer to the transmit block which allows the two blocks to communicate, the file to which the received data should be stored, how many samples of each PRI should be recorded to file and the number of samples to be ignored in each PRI until the receiver should begin recording. Upon instantiation, the receiver checks the system state. If the system is in the WAIT state it simply ignores all input samples. If the systems is in the SYNC, PREAMBLE or TRANSMIT state, this means the transmitter has begun transmitting. The GNU Radio package includes by default several building blocks for signal and information processing. The programmer has the
option to custom design signal processing blocks according to his/her specifications. Further details about signal processing steps in SDR have been explained in [7]. In GNU Radio signal processing blocks are built using a combination of Python code for high level abstraction, GUI and other non performance-critical functions, while performance critical signal processing blocks are written in C++. Swig is used as an interface between C++ and Python. These custom designed blocks can then be accessed from GRC as any other default signal processing block with the help of XML.
Fig.6: Block diagram of software-defined radar in operational mode
The blocks GNU Radio package includes by default several building for signal and information processing. Some of the implemented blocks are FFT blocks, FIR, IIR and Hilbert filters, automatic gain control blocks and various modulation and demodulation blocks. Apart from these signal processing blocks, various sink and source blocks such as signal generators, noise generators, audio source and sink and UDP (User Datagram Protocol) source and sink are also pre-defined. The programmer has the option to custom design signal processing blocks according to his/her specifications. The main signal processing blocks used in synthetic aperture radar (SDR) are focusing, speckle filtering, radiometric calibration and normalization. Doppler Centroid estimation requires the autocorrelation function which is coded in C++ and then run in Python using SWIG2 as the interface. Range compression and azimuth compression are done by performing FFT and inverse FFT on the received data. As the FFT blocks are in-built in GNU Radio software, the compression blocks can be implemented as a graph. Speckle filtering is done so as to remove the random statistical fluctuations associated with the radar reflectivity of each pixel of the image. This is achieved by using adaptive filters based on minimum square error or maximum aposteriori (MAP). The received pulses are classified using Support Vector Machine algorithm. This algorithm should be insensitive to noise effects and any small changes in the data. The data collected at one time and location with a certain signal to noise ratio can be used to classify pulses at another time and location with a possibly differing SNR [5]. The radiometric calibration of the SAR image is done by considering the radar equation law for corrections in the scattering area (A), the antenna gain pattern (G 2) and the range spread loss (R3). Radiometric calibration can be implemented in SDR by writing a C++ program. The variations in the backscatter energy due to the radar look angle and swath width are normalized by performing a modified cosine correction. VII.
EXPERIMENTS
When the signal is received by the receiver antenna, it is stored as a log file in GNU Radio, which is of type gr_complex. This log file can be plotted with the help of the
folder MATPLOTLIB that isthe installed along with theafter GNU radio software as it contains necessary commands converting it into matrix or any other data format as required. After the received signal is obtained, subsequent signal processing is done and calibration techniques are applied to get an idea of the information that has been transmitted. The functions used for plotting are gr_plot_fft, gr_plot_psd and gr_plot_iq to get the power spectral density, fast fourier transform and in-phase and quadrature phase components of the log file respectively. The following experiment describes how an SDR can be used effectively for plotting the data received by an SAR and hence can be used for analysing the type of signal that was transmitted. Fig 7 shows the typical experimental setup with PC and USRP kit. For SAR, the antenna in the experimental setup should be moved and since we use USRP, the experimental setup can be moved more freely. The connection between USRP and System can be established by wireless means. For this experiment, frequency of 2.5GHz is considered to be suitable and hence RFX2400 (daughter board) is used since it operates in the frequency range 2.25-2.9GHz. A chirp signal of width 32MHz is transmitted with a pulse repetition frequency (PRF) of 100Hz. The same experiment can also be performed at different PRFs
Fig 7: Typical Experimental Setup
A Pulse period of 5 µs and waveform amplitude of 15mv was used. Initially the experiment is performed within a small radius and using the USRP kit the signal received on reflection from the object is stored as a log file in the system. In the first case, reading was taken in a free environment. In the second case, a human body is brought into the test area and the subsequent readings are also taken and stored. And finally a metal object was placed in the test
area and readings were taken. The plot for the first case and third case is given in figure 8 and figure 9 respectively. The x-axis and y-axis are sample number and amplitude respectively for the I&Q plot. These files are stored as ‘.dat’ files and it is plotted with the help of MATPLOTLIB files in GRC to get various plots such as power spectral density, fast Fourier transform and in-phase and quadrature phase components.
Fig.9: I&Q and FFT plot of the reflected signal from the metal surface.
Fig.8 I&Q and FFT plot of the reflected signal in free space.
From the Fig.8 and 9 we observe a change in the Qcomponent. This change in the amplitude in Fig.9 helps to estimate the presence of an object. The variation can be seen distinctly in the FFT plots. In the similar fashion different objects can be placed at some distance from the antenna in the USRP kit and change in the amplitudes can be recorded. The data collected in the above fashion acts as a database. When this setup is used for the detection of target or enemies in practical situations, the database will be of good use. VIII.
COMPARISON WITH CONVENTIONAL RADAR
On comparing the proposed idea of implementing Radar using SDR with the normal conventional Radar, the following disparities were found and they are tabulated below (Table 1). S.No
Properties
ConventionalRadar
1.
Effect of Amplifier characteristics
Possibility of non-linearities in the output of the amplifier is high
RadarusingSDR
2.
Local oscillator
The changes in frequency stability and phase shift leads to spurious output signals in the output of the mixer
3.
Waveform Adaptation
The hardware design can be implemented The type of waveform can be changed as and for only one type of waveform in particular when necessary.
4.
Portability
This type of radars are mostly suitable for static application.
5.
Application
The hardware radar is designed and deployed for a specific application
6.
Reconfigurability
Configuration is one time
7.
Upgradation cost
High, due to cost incurred for replacing several components.
The problem of non-linearity does not arise because amplification is done by scaling in a software This problem is avoided here completely. Signal generation is digitally done in PC.
Since we go in for implementation using USRP and PC, the experimental setup can be placed anywhere and is mobile in nature. It can be used for versatile applications by modifying the software module. On-the-fly configuration is possible Drastically Low
Table 1: Comparison between Conventional Radars and Radars using SDR.
IX.C ONCLUSION The use of software to synthesis radar leads one to ask whether cognitive radar can be constructed that is capable of sensing the surrounding wireless environment and user
communications and computing needs and acting to meet those needs. The additional blocks have the functions of “sniffing” the radar environment, providing for dynamic spectrum management, and controlling the level of transmitted power.
However, considering that cognitive radio functions are inherent to physical (PHY) and medium access control (MAC) layer operations, the resulting “embedded cognition” requires a robust model and framework which is capability of operating within the computing constraints available in current wireless platforms. In addition, traditional machine learning techniques require significant computational resources, which could limit the utility of a cognitive radio. These challenges are to be overcome. In this paper, a novel approach to implement the synthetic aperture radar using Software defined radio (SDR) concept has been proposed. The blocks in an SAR are defined in the GNU Radio software. The RF front end is implemented by using the minimum hardware - USRP (Universal Software Radio Peripheral). The digital data (received data) from the USRP is sent to the host CPU, where it is processed to obtain the radar image. By using software modules, more flexibility and reconfigurability of the radar design can be obtained. The testing and configuration analysis of the implemented prototype is in progress. The research is expected to extend to a pint of realizing cognitive radar REFERENCES [1] [2] [3]
[4] [5] [6] [7]
[8]
Lee K. Patton , A GNU Radio Based Software-Defined Radar , Master’s Thesis, Wright University, April 2007. Synthetic Aperture Radar - Land Applications, Tutorial: Part I Background and Theory, ESA. Danilo Valerio, Open Source Software-Defined Radio: A survey on GNU radio and its application , FTW Technical Report, August 2008. Matt Ettus, USRP User’s and Developer’s Guide , Ettus Research LLC. Gregory P. Noone, Radar Pulse Classification using Support Vector Machines. B. P. Lathi, Signal Processing and Linear Systems , BerkeleyCambridge Press, 1998. Behrouz Farhang-Boroujeny, Signal Processing Techniques for Software Radios, Department of Electrical and Computer Engineering, University of Utah Eric Blossom, “Gnu Radi o: GnuRa dioHardware”, http://www.comsec.com/.
