Random number generator based on intramuscular electromyography to secure wireless networks of electronic implants

Adami Peracchia, Dante

Random number generator based on intramuscular electromyography to secure wireless networks of electronic implants

Citation

Adami Peracchia, D. Random number generator based on intramuscular electromyography to secure wireless networks of electronic implants. 2022. handle: http://hdl.handle.net/10230/54581

This citation was generated automatically.

Abstract

Miniaturization is one of the biggest challenges faced by researchers and engineers working on wireless electronic implants. In this context, the Biomedical Electron- ics Research Group (BERG) of the Pompeu Fabra University (UPF) works on a new method to perform intramuscular electrical stimulation and electromyography (EMG) through distributed wireless networks of miniaturized implants that bidirec- tionally communicate with wearable external units. In the future, the communica- tions between the wireless networks of devices and the wearable external units will have to be encrypted to prevent attacks that could jeopardize the health and data of the users. Nowadays, random sequences of bits known as keys are used to encrypt and decrypt signals. These keys are typically generated with Pseudo Random Num- ber Generators (PRNG) which use as sources of entropy combinations of the states and processes of the hardware. In the envisioned scenario, the limited hardware resources hamper the generation of secure and random sequences. Several authors have proposed the use of different biosignals as effective sources of entropy to gen- erate True Random Number Generators (TRNG). In this work, a TRNG based on intramuscular EMG is proposed, hypothesizing that it will generate more random sequences with higher efficiency than a PRNG of a device with limited hardware resources. Furthermore, the effect of different contraction levels and neuromuscular diseases on the EMG and the generated strings is studied to analyze all the sce- narios where such TRNG can be implemented. Results show the capability of the proposed algorithm to generate high-performance random sequences of bits from in- tramuscular EMG with less computational cost than a PRNG of an electronic device with limited hardware. The performance of the TRNG is not affected by the used muscle as source of EMG or its level of effort. However, results show differences in performance when using data from patients suffering from neuropathic diseases.