Exercise 4: Solution Development

Proposal:

Using SDR equipment provided by the Airforce Research Laboratory and the Wright Brother's Institute, students shall develop a software defined radio system with improved adaptability. By implementing machine learning to previously made SDR algorithms, we will produce an improved spectrum sensing radio system, as well as a faster and more effective way to sense and utilize signals.

General Design Goals

• Use Python to develop algorithms

• Using hardware:

  • Laptop

  • NI-USRP 2901

  • USB 3.0 Cable Type A to B

  • 30 dB SMA Attenuator

  • Multiple Antennas

• Low-Cost:

  • Use provided hardware and free software

Specific Design Goals

• Use Open-Source Tools with USRP Hardware

  • Go to learn.ni.com > click "View Library" > Under Browse search "Software Defined Radio" > Begin Learning Path (Using Open-Source Tools with USRP hardware for SDR Applications)

    • Get familiar with all things SDR

    • This learning path explains how to set up the SDR with the UHD Driver and GNU radio

  • Go to Ettus Knowledge Base and click on Application Notes

    • To install UHD Driver:

      • AN-445: building driver from Linux, will show GitHub for the driver, under host folder, high level doc for how to build

    • Click AN-088 for tuning tips and follow instructions to optimize performance of SDR

  • Once familiarized enough with GNU Radio and the Open-Source Tools, students shall implement machine learning to give the SDR more adaptability (will get more specific as time goes on...)

• LabVIEW

  • Download NI-USRP Driver
    NOTE: LabVIEW should be downloaded before downloading the driver

    • This will set up the NI-USRP Driver in LabVIEW

  • Download and install LabVIEW Modulation Toolkit

    • Gives several communications examples on LabVIEW

  • To find and use NI-USRP Palette:

    • Open Project > Right click > Instruments I/O > Instr Drivers > NI-USRP

  • Use of LabVIEW to receive and transmit signals

  • Explore different examples on the driver to become familiar with the software

• Design Goal Specifications: USRP 2901 Specifications (Link for a list of USRP 2901 Specs)

  • Bandwidth limited to 56 MHz

  • NI-USRP Sampling Rates:

    • Frequency Range: 70 MHz - 6 GHz

    • Master Clock Rate (MCR): Programmable between 1 MHz and 61.44 MHz (30.76 when operating in 2x2 mode)

      • Whatever sample rate is chosen must be supported by the MCR

      • Decimation rates between 1 and 1024

      • We set sample rate and driver attempts to match MCR and Decimation Rate to our sample rate

    • Sample at highest frequency component, no need for Nyquist Theorem since there are 2 ADCs

      • Sometimes requires higher sampling

Final Solution State

• SDR with fully functioning self-learning capabilities

• Effective spectrum sensing

• Higher knowledge of SDRs and how they work

• Easily access internal values and states from anywhere within range

In Class: Solutions

TL;DR:

keyword is shall. "Students shall..."

TPU:

Could be used for Senior Project, good machine learning library.

Coral USB accelerator, unavailable at the time.

Machines for this in B219, to be moved into another room soon, and hopefully will be introduced to it eventually in the semester.

Quantum Computing:

Link