Ali Najafi

Measuring Middle Ear Function Using a Smartphone-based Tympanometer.
The primary tool for detecting numerous ear conditions is a machine called tympanometer. However, such machines cost thousands of dollars and are typically bulky and inflexible for use in resource-limited settings. We developed a smartphone-based prototype which costs less than $20. Our prototype uses a microphone, speaker, pressure sensor and pressure pump to change the pressure in the ear canal which are air-sealed together with a piping structure. The microphone and speaker are connected to a smartphone which transmits acoustic signal and measures acoustic immittance in the ear canal. Our prototype functionality matches the capabilities of a clinic-grade tympanometer. We are currently validating the clinical efficacy of this tool in several clinical settings, and are planning to run a deployment in audiology clinics in resource-constrained environments.

Building a smart speaker for medical research.
We built a custom smart-speaker for real-time respiratory pattern and cough monitoring with the goal of early detection of infectious diseases. It transmits an ultrasound (>20 KHz) signal, processes the received 8-microphone array signal reflected from the human body on the Raspberry Pi to extract the real-time respiratory patterns and sends them to the cloud.

Wireless Steerable Vision for Live Insects and Insect-scale Robots.
Existing power-autonomous insect-scale robots do not have the ability to capture vision data given their power, weight and size requirements. We designed a fully wireless mechanically steerable vision system in a form factor compatible with small robots and live insects. Our system can mechanically steer a camera over a 60 degree angular range using a millimeter scale actuator that is wirelessly controlled from a smartphone. The electronics and actuator weigh 248 mg and can stream 160×120 monochrome video at 1–5 frames per second (fps) to a Bluetooth radio from distances of up to 120 m away. Using thissystem, we built the smallest robot (1.6×2cm) that supports a wireless steerable camera.

TinySDR: Low-Power SDR Platform for Over-the-Air Programmable IoT Testbeds.
We built a low-power, cheap and small form-factor Software-Defined Radio (SDR) platform suitable for IoT applications. TinySDR supports over the air update for physical layer on FPGA and MAC layer on microcontroller and achieves 30uW of sleep power, 10000 times lower than other SDR platforms. It supports 4~MHz of bandwidth on sub-GHz and 2.4GHz frequencies and provides sensor interfaces. To showcase the functionality of this platform, we implemented chirp spread spectrum (CSS) modulator/demodulator and also Bluetooth beacons and achieved these standard's sensitivities.

NetScatter: Enabling Large-Scale Backscatter Networks.
We developed the first wireless protocol that scales to hundreds of concurrent transmissions from backscatter devices and works below the noise floor. Our coding technique combines Chirp Spread Spectrum (CSS) and ON-OFF keying to utilize unused frequency bins. We deployed our design using a testbed of backscatter hardware and showed that our protocol scales to concurrent transmissions from 256 devices using a bandwidth of only 500 kHz. We implemented the baseband of backscattering devices using IGLOO Nano FPGA and MSP430 microcontroller. The access point was implemented using X-300 USRP and MATLAB.

LoRa Backscatter: Enabling The Vision of Ubiquitous Connectivity.
We developed the first communication technology that provides reliable and long-range communication at tens of microwatts of power as well as cost less than a dime. To achieve this, we implemented Chirp Spread Spectrum (CSS) modulation on an ultra-low power tag. We built a custom PCB prototype tag. The baseband section of the tag is implemented on an IGLOO Nano FPGA and generates switching signal for the RF switch which is connected to the antenna. We showed that with our system and having Semtech LoRa SX1276 as receiver, we can achieve several hundred meters of wireless communication range.

Wideband Self-Interference Cancellation for Full-Duplex Transceivers.
We designed and fabricated a transceiver front-end IC with a dual-injection path self-interference (SI) cancellation for wideband wireless full-duplex networks. The SI cancellation circuitry is implemented using: 1) feedforward cancellation path between the transmitter output and the receiver input; 2) baseband cancellation path injecting at the RX output; 3) phase noise cancellation method that suppresses the reciprocal mixing products. A prototype was fabricated using 40nm TSMC CMOS technology achieving state of the art cancellation depth and bandwidth and presented the measurement results.

Precision Wideband Quadrature Generation for Millimeter-Wave Communication Systems.
Utilizing a direct-conversion architecture to build a mm-wave transceiver requires in-phase and quadrature-phase signals that are highly phase accurate (90°), with a minimal amplitude mismatch. This is challenging because of an increased sensitivity of quadrature phase and gain accuracy for a given mismatch as the frequency rises. To address this, we designed and fabricated an integrated two-stage polyphase filter with feedback control for quadrature local oscillator generation at millimeter-wave frequencies using 28nm LP CMOS process and achieved state-of-the-art results.

Regenerative Breaking: Recovering Stored Energy from Inactive Voltage Domains for Energy-efficient Systems-on Chip.
We presented an approach to leverage existing voltage regulators to recover the otherwise wastedenergy of the significant output capacitance of the disabled voltage-domain back into the supplyusing a low-overhead all-digital run-time control system. We have implemented this technique on65nm CMOS using synthesis, auto-place and route design-flow and show that we can recover over90% of the stored energy across a range of operating system voltages.