Abstract

The edge devices in an emerging Internet-of-Things (IoT) environment require comprehensive security measures that are within the power budget for ubiquitous computing. In this paper, a transmitter identification scheme consisting of a lightweight Bayesian neural network (BNN)-based classifier using raw time-domain data is presented. Evaluation is performed with data obtained in schematic-level simulation of high-efficiency CMOS power amplifier designs using a 65 nm process design kit (PDK). The Bayesian neural networks achieve 89.5% accuracy on the task of classifying six transmitters. Moreover, the BNN classifier is implemented on field-programmable gate array (FPGA) with parallel pseudo-Gaussian random number generators to achieve a throughput of more than 340,000 classifications per second, with average energy consumption for each classification task of 0.548 μJ. This low-power system enables comprehensive security for energy-constrained IoT devices and sensors.

Highlights

  • W IRELESS connection increases the communication efficiency and information sharing through ubiquitous sensing and computing platforms [1], [2]

  • Various features have been utilized for radio fingerprinting [9], including intellectual property (IP)/medium access control (MAC) addresses, radio signal strength (RSS), and channel state information (CSI), which are subject to environmental changes

  • The number of PEs in the PE array is equal to the number of neurons in the layer, which consumes an equal number of DSPs in the field-programmable gate array (FPGA) so that the number of clock cycles it takes to get the matrix multiplication results is equal to the number of inputs plus one MAC’s latency, which effectively accelerates the forwarding process of the neural network with efficient hardware resource utilization

Read more

Summary

INTRODUCTION

W IRELESS connection increases the communication efficiency and information sharing through ubiquitous sensing and computing platforms [1], [2]. To show the distribution of features within the alternative dataset after passing through the low-IF receiver behavioral model, the FFT was taken for an overall LNA and channel gain of −20 dB, an IIP2 of 60 dBm, and an IIP3 of 10 dBm across transmitters at 27◦ C and nominal VDD without adding AWGN We observed that the accuracy at SNR = 45dB did not improve compared to SNR = 40dB This is considered as the result of the model’s generalization with high SNRs as we trained the model with mixed-SNR signals and the accuracy does not progress with high SNRs. we investigated the impact of environmentally-induced variations in receiver nonlinearity on the inference accuracy of a trained classifier for both IIP2 and IIP3 with random SNRs and overall channel and LNA gains of −20 and −40 dB. This BNN model was deployed on FPGA using the trained parameters for performance verification

FPGA INFERENCE MODULE FOR BNN
Findings
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call