Stochastic Computing based AI System for Mobile Devices

Stochastic Computing based AI System for Mobile Devices

Table of Contents


In this paper, we present a stochastic computing-based AI system for mobile devices. As technology in AI advances, more complex computations are required. In the case of mobile devices, it is hard to accommodate the entire computations due to the power and area limitations of an embedded system. As stochastic computing replaces the complex computations with simple computations, mobile devices are available to include the AI system. In order to verify our design, the embedded AI system including stochastic computing is implemented on a field-programmable gate array (FPGA), and we successfully demonstrated the feasibility of the proposal.

  • IEEE Keywords

    • Artificial intelligence ,
    • Neurons ,
    • Mobile handsets ,
    • Field programmable gate arrays ,
    • Embedded systems ,
    • Generators ,
    • Conferences


A technology advance across artificial intelligence (AI) increases the computation complexity. The complex computations help the AI system to achieve high accuracy, but also consume a large number of power and area. Especially, the embedded system has limitations about area and power due to the characteristic of embedded devices [1]. For that reason, the embedded system is hard to contain many multiplications for computation in a restricted environment. In order to overcome the limitation, we need a specific computation module to reduce logic area and power. Stochastic computing is a low-complexity computation module. Stochastic computing consists of a bit-stream to express probability in a ratio of 0 and 1 [2]. The stochastic bit-stream represents the number of 1 as the size of the data. Stochastic computing performs multiplication with logic operations to reduce computation size. According to the smaller computation logic size, the system requires less power consumption. For that reason, stochastic computing is available to be used in various fields to conduct many operations such as image processing, filter, and machine learning algorithm [3-5]. When the AI system includes stochastic computing, the system is minimized than including the multiplications without stochastic computing [6]. Therefore, stochastic computing is effective to the embedded AI system in terms of area and power consumption.

In this paper, we propose a stochastic computing-based AI system for mobile devices. The proposed stochastic computing performs the simple AND operation instead of the complex square operation. In order to verify the functionality of stochastic computing, we demonstrated our design with the embedded AI system by implementing on a field-programmable gate array (FPGA).


Fig. 2 shows the architecture of the stochastic computing. The stochastic computing consists of stream generator, computation, and stream to binary logics. By the characteristic of the stochastic computing, the stochastic computing needs to convert binary data into a stochastic bit-stream. The stream generator, which focuses to express the value of data stochastically, consists of random number generators and comparator. The random number generators use linear feedback shift registers (LFSRs) for hardware design and transfer the 8-bit LFSRs to the comparator. The comparator consists of dual LFSR memories and compare unit to generate bit-streams that the length is 2 power 8. Each LFSR memory has 8 widths and 128 depth to save the 8-bit LFSR from the random number generators. In order to generate the bit-stream, the input data is compared with both LFSR memories as the comparator receives input data. When the value of data is larger than the LFSR, the one bit of the bit-stream sets 1. After comparing between the data and the LFSRs, the computation logic transmits the generated bit-streams. The computation logic operates the bit-streams by AND operation to substitute for multiplications when the send signal sets high state. After that, the computation result is transferred to the stream to binary logic to convert to binary data. When the stream to binary logic receives the complete signal from the computation logic, the stream to binary logic counts 1 in the operated bit-stream from the computation logic to convert from the bit-stream to the value of binary data.


In this paper, we propose a stochastic computing-based AI system for mobile devices. In the embedded system which requires complex operations, stochastic computing helps to replace the operations simple. Also, stochastic computing enables to reduce the area and power consumption. By adopting stochastic computing in the AI processor for face recognition, we implemented the proposed design on the FPGA. After demonstrating the functionality of stochastic computing, we verified the feasibility of stochastic computing for the embedded system.

About KSRA

The Kavian Scientific Research Association (KSRA) is a non-profit research organization to provide research / educational services in December 2013. The members of the community had formed a virtual group on the Viber social network. The core of the Kavian Scientific Association was formed with these members as founders. These individuals, led by Professor Siavosh Kaviani, decided to launch a scientific / research association with an emphasis on education.

KSRA research association, as a non-profit research firm, is committed to providing research services in the field of knowledge. The main beneficiaries of this association are public or private knowledge-based companies, students, researchers, researchers, professors, universities, and industrial and semi-industrial centers around the world.

Our main services Based on Education for all Spectrum people in the world. We want to make an integration between researches and educations. We believe education is the main right of Human beings. So our services should be concentrated on inclusive education.

The KSRA team partners with local under-served communities around the world to improve the access to and quality of knowledge based on education, amplify and augment learning programs where they exist, and create new opportunities for e-learning where traditional education systems are lacking or non-existent.

FULL Paper PDF file:

Stochastic Computing based AI System for Mobile Devices



S. Y. Jang, Y. H. Yoon and S. E. Lee,




Stochastic Computing based AI System for Mobile Devices

Publish in

2020 IEEE International Conference on Consumer Electronics (ICCE), Las Vegas, NV, USA, 2020, pp. 1-2,



PDF reference and original file: Click here

+ posts

Somayeh Nosrati was born in 1982 in Tehran. She holds a Master's degree in artificial intelligence from Khatam University of Tehran.

Website | + posts

Professor Siavosh Kaviani was born in 1961 in Tehran. He had a professorship. He holds a Ph.D. in Software Engineering from the QL University of Software Development Methodology and an honorary Ph.D. from the University of Chelsea.

Website | + posts

Nasim Gazerani was born in 1983 in Arak. She holds a Master's degree in Software Engineering from UM University of Malaysia.