Performance Characterization of USRPs

Abstract

Software Defined Radios (SDRs) are systems in which components such as filters, modulators, demodulators, etc., typically implemented in hardware are instead implemented in software. Thus, SDR systems shift signal processing from analog to digital domain. The signal processing is performed in reconfigurable devices like General Purpose Processors (GPP) or Field Programmable Gate Arrays (FPGA). The resulting flexibility of SDR systems brings in many advantages to wireless communications such as improved interoperability, adaptation capability and more future-proof hardware.

Various prototype/laboratory friendly hardware are available such as Universal Software Radio Peripherals (USRPs), developed by Ettus Research, which in combination with software interfaces such as GNU Radio can be used to design and implement a Software Defined Radio(SDR) system. This has in turn enabled numerous research opportunities and advancements in field of wireless communication. However one major drawback of these simple USRP devices is that they are un-calibrated in terms of power or voltage and hence give the results in relative terms/counts. This poses a disadvantage in real world scenarios where parameters such as power measured in milliwatts or dBm, are used to design, implement, and test devices and systems and to collect and interpret experimental results. Lack of connectivity with these metrics leads to the USRP devices being confined to proof of concept implementation and demonstrations, and limits their utility for experimentation.

In this thesis an attempt has been made to calibrate the USRPN210 (with WBX daughter board) devices and derive some calibration factors that would help relate the experimental results obtained from the USRP and GNU Radio combination to real world metrics. This also makes the result-set clearer and easily understandable. Several experiments were conducted to understand and characterize the performance of the USRP under different conditions. Experiments such as determination of 1-dB compression point, Third order intercept point helped understand the linearity range of the device under transmitter and receiver operations. The 3rd order input intercept point for USRPN210 device with WBX board with calibration was found to be around 0.25 dBm, which is closer to the expected value of 0 dBm at 400 MHz. Amplitude and frequency stability tests over the operational frequency range of the daughter board helped to note if change in frequency produced any variation in the received or transmitted power. Also variation with respect to gain helped derive a reference table for transmitted power, which could be used in for future experiments. Conducting these experiments over a sample of 10 USRPs helped in obtaining standardized values and benchmarking them. The factors obtained were also used in some experiments like path loss modeling, position location estimation in order to determine their effectiveness and impact on such applications.

Although, these factors obtained may not be directly applicable to all USRP devices, as in any analog circuitry, and also since they are conducted relative to some specific applications such as the predefined spectrum analyzer utility of GNU Radio (UHD_FFT.grc), they do help to understand the basis of derivation of calibration metrics resulting from the various experiments and also provide a platform for future work in this area to obtain more uniform and globally acceptable results. All photos by the author unless cited, 2014.