DustNet: A Wireless Network of Ultrasonic Neural Implants

TL;DR

DustNet addresses the challenge of chronic, high-bandwidth neural recordings from deeply seated peripheral nerves without percutaneous wires by leveraging a wireless, ultrasonically powered network of up to eight implants. It introduces a TDMA backscatter scheme with 16-level amplitude modulation to achieve a total data rate of 400 kilobits per second (50 kilobits per second per implant) over a single ultrasound link at a 2 megahertz carrier, while consuming approximately seven microwatts per implant. The system is implemented in a 28-nanometer CMOS DustNet IC (0.43 square millimeters) featuring power management, an ultrasound interface, a neural recording front end, and a digital backend, and is validated at depths exceeding 90 millimeters in tissue-mimicking media with a measured bit-error rate as low as 1.89e-5. Compared with prior work, DustNet demonstrates the highest reported simultaneous US-powered implants, deep operation capability, and the most spectrally efficient ultrasound-based neural interfacing to date, promising practical long-term peripheral nerve control for prosthetic devices.

Abstract

Spatially distributed peripheral nerve recordings can be used to reconstruct motor intention and improve natural control of prosthetics in patients with limb deficiencies. However, many existing clinical solutions rely on percutaneous wires to access peripheral nerves; these sites are prone to infection and electrode degradation, preventing chronic use. To enable longterm recording of deeply-seated peripheral nerves, this paper presents DustNet: a network of ultrasonically-powered neural recording implants capable of supporting up to 8 simultaneously recording nodes over a single ultrasound link. To enable high throughput multi-implant communication, DustNet implements a time-division multiple-access (TDMA) protocol with up to 16-level amplitude modulation of the ultrasound backscatter that achieves up to 4x higher data rates than traditional on-off keying methods. Each neural implant consists of a 0.7x0.7x0.7 mm^3 piezoceramic transducer, a 10 nF off-chip capacitor, and an IC mounted on a flexible PCB. The implant IC was fabricated in a 28nm CMOS process and occupies an area of 0.43 mm^2. System functionality was verified within FDA power limits at 90mm depth, achieving a maximum data rate of 400 kb/s at 2 MHz ultrasound carrier frequency, with each implant transmitting uplink data at 50 kb/s and dissipating just 7 μW of power

Figures (15)

• Figure 1: (a) Conceptual diagram of the proposed DustNet wireless neural recording platform with 8 implanted sensor nodes. (b) Implant mock-up diagram and assembly with a US dime for reference.
• Figure 2: Timing diagram for pulse-echo communication using CDMA and TDMA protocols. During uplink, the maximum achievable data rate for both protocols is identical, but TDMA is more robust to inter-implant mismatch and allows implants to charge without sacrificing channel bandwidth.
• Figure 3: (a) Typical implant piezo impedance as a function of frequency. At resonance ($f_s$ and $f_p$), the piezo impedance is purely resistive. (b) Piezo electrical model at resonance. The resistance at the series resonance is significantly lower than the resistance at the parallel resonance.
• Figure 4: Multi-level ASK modulation: (a) Using a linear R-DAC and (b) using a linear I-DAC.
• Figure 5: Complete timing diagram of the DustNet communication protocol. In Config Mode, the external transducer transmits pulses that encode link parameters and a target implant ID. Implants matching the target ID backscatter the ID to acknowledge successful configuration. A transmitted target ID of 0 signals implants to transition to Uplink Mode, which implements the TDMA protocol for data transmission.