Masaya Fujisawa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Masaya Fujisawa has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Kazunori Uruma
Department of Computer Science, Kogakuin University, Shinjuku, Tokyo, Japan
2 shared publications - Norisato Suga
College of Engineering, Shibaura Institute of Technology, Koto, Tokyo, Japan
2 shared publications - Yoshihiko Tsuchiya
Department of Information and Computer Technology, Tokyo University of Science, Tokyo, Japan
2 shared publications - Ahmad Akmal Aminuddin Mohd Kamal
Department of Information and Computer Technology, Tokyo University of Science, Tokyo, Japan
1 shared publication - Hiroyoshi Morita
Dept. of Computer and Network Engineering, Graduate School of Informatics and Engineering, The University of Electro-Communications
1 shared publication Masamune Okada
Graduate School of Engineering, Tokyo University of Science, Tokyo, Japan
1 shared publication
Latest Publications
- Open Access
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 49961-49974
Integration of CSI and Camera Image for Indoor Wireless Localization
Research in integrated sensing and communications is advancing toward next-generation systems beyond 5G and 6G, with significant progress in highly accurate indoor localization methods, particularly those leveraging channel state information (CSI) and deep learning (DL) techniques. However, most existing approaches assume access points equipped with a large number of antennas, which limits their applicability in current communication systems that typically use fewer antennas. Furthermore, DL-based localization methods often require extensive training datasets, posing a significant challenge in terms of the cost of generating training samples. This study addresses these challenges by proposing a camera-assisted localization framework designed to reduce both the number of required antennas and the cost associated with generating training samples. The proposed method integrates object detection from camera images to identify users among multiple detected objects and combines the estimated user coordinates from images with CSI data to achieve precise localization. This approach achieves high localization accuracy with a limited number of antennas and training samples. Furthermore, to minimize the effort required to acquire labeled data, the framework incorporates a semi-supervised learning strategy that uses a small set of labeled data together with a larger set of unlabeled data. This method enhances localization accuracy while reducing the cost of training data acquisition. Numerical simulations demonstrate that the proposed method achieves superior localization accuracy with fewer antennas and labeled data compared with conventional methods that rely solely on CSI.
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 49961-49974
Open AccessIntegration of CSI and Camera Image for Indoor Wireless Localization
Research in integrated sensing and communications is advancing toward next-generation systems beyond 5G and 6G, with significant progress in highly accurate indoor localization methods, particularly those leveraging channel state information (CSI) and deep learning (DL) techniques. However, most existing approaches assume access points equipped with a large number of antennas, which limits their applicability in current communication systems that typically use fewer antennas. Furthermore, DL-based localization methods often require extensive training datasets, posing a significant challenge in terms of the cost of generating training samples. This study addresses these challenges by proposing a camera-assisted localization framework designed to reduce both the number of required antennas and the cost associated with generating training samples. The proposed method integrates object detection from camera images to identify users among multiple detected objects and combines the estimated user coordinates from images with CSI data to achieve precise localization. This approach achieves high localization accuracy with a limited number of antennas and training samples. Furthermore, to minimize the effort required to acquire labeled data, the framework incorporates a semi-supervised learning strategy that uses a small set of labeled data together with a larger set of unlabeled data. This method enhances localization accuracy while reducing the cost of training data acquisition. Numerical simulations demonstrate that the proposed method achieves superior localization accuracy with fewer antennas and labeled data compared with conventional methods that rely solely on CSI.
- Open Access
- 21 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 73625-73651
Privacy-Preserving Keyword Search With Access Control for Secret Sharing-Based Data Outsourcing
Searchable encryption, also known as secure search, is a technology that enables search operations on encrypted data while maintaining its confidentiality. Extensive research has been conducted on searchable encryption utilizing public-key and symmetric encryptions. However, public-key encryption involves significant computational expenses and is deemed inefficient for querying large databases, particularly in cloud settings. Hence, our attention is on searchable encryption using a secret sharing scheme (also known as searchable secret sharing), which is recognized for its minimal computational complexity. A secret sharing scheme is a method to transform a secret input into several distinct values known as shares. Kamal et al. introduced a simple searchable encryption utilizing secret sharing schemes, yet did not include user access control capabilities. In this study, we introduce a secure search method with user access control employing a secure computation based on a (k, n) threshold secret sharing scheme, where every data stored in the cloud has an owner, and owners grant access to users at their discretion.We assume a client-server model to perform a secure computation between the owner, the player, and the n ≥ k cloud servers. Furthermore, we provide an in-depth analysis of the security aspects of our proposed distribution, query generation, and search processes, demonstrating that our approach is resilient against an honest-but-curious adversary with information from up to k−1 computing servers. Moreover, we also include an improved and efficient method by using a (n, n) additive secret sharing scheme when assuming n = k computing servers. We provide discussion comparing both our proposed methods in terms of their computation and communication costs. Finally, we also include a detailed performance analysis using Python, including a comparison with conventional secret sharing-based methods.
- 21 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 73625-73651
Open AccessPrivacy-Preserving Keyword Search With Access Control for Secret Sharing-Based Data Outsourcing
Searchable encryption, also known as secure search, is a technology that enables search operations on encrypted data while maintaining its confidentiality. Extensive research has been conducted on searchable encryption utilizing public-key and symmetric encryptions. However, public-key encryption involves significant computational expenses and is deemed inefficient for querying large databases, particularly in cloud settings. Hence, our attention is on searchable encryption using a secret sharing scheme (also known as searchable secret sharing), which is recognized for its minimal computational complexity. A secret sharing scheme is a method to transform a secret input into several distinct values known as shares. Kamal et al. introduced a simple searchable encryption utilizing secret sharing schemes, yet did not include user access control capabilities. In this study, we introduce a secure search method with user access control employing a secure computation based on a (k, n) threshold secret sharing scheme, where every data stored in the cloud has an owner, and owners grant access to users at their discretion.We assume a client-server model to perform a secure computation between the owner, the player, and the n ≥ k cloud servers. Furthermore, we provide an in-depth analysis of the security aspects of our proposed distribution, query generation, and search processes, demonstrating that our approach is resilient against an honest-but-curious adversary with information from up to k−1 computing servers. Moreover, we also include an improved and efficient method by using a (n, n) additive secret sharing scheme when assuming n = k computing servers. We provide discussion comparing both our proposed methods in terms of their computation and communication costs. Finally, we also include a detailed performance analysis using Python, including a comparison with conventional secret sharing-based methods.
- Open Access
- 30 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 2829-2843
Active Element Arrangement for Deep Learning-Based CSI Prediction in IRS-Assisted Systems
Recently, intelligent reflecting surface (IRS) components have garnered attention as a technology for advancing next-generation wireless communications. For effective IRS control with minimized overhead, a scheme is needed for estimating partial channels using active elements. Given that the number of active elements affects production costs and energy consumption, various methods have been proposed for predicting channels with the fewest elements. However, despite the significant impact of active element arrangement on prediction accuracy, as corroborated in this study, there has been limited discussion on this aspect, especially in deep learning (DL)-based prediction methods. In this study, we conducted experimental investigations to explore the conditions under which an active element arrangement resulting in accurate DL based predictions. Subsequently, we developed a novel active element arrangement based on three policies extracted from the simulation results. Through numerical experiments, we demonstrated that the proposed arrangement facilitates accurate prediction regardless of the number of active elements.
- 30 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 2829-2843
Open AccessActive Element Arrangement for Deep Learning-Based CSI Prediction in IRS-Assisted Systems
Recently, intelligent reflecting surface (IRS) components have garnered attention as a technology for advancing next-generation wireless communications. For effective IRS control with minimized overhead, a scheme is needed for estimating partial channels using active elements. Given that the number of active elements affects production costs and energy consumption, various methods have been proposed for predicting channels with the fewest elements. However, despite the significant impact of active element arrangement on prediction accuracy, as corroborated in this study, there has been limited discussion on this aspect, especially in deep learning (DL)-based prediction methods. In this study, we conducted experimental investigations to explore the conditions under which an active element arrangement resulting in accurate DL based predictions. Subsequently, we developed a novel active element arrangement based on three policies extracted from the simulation results. Through numerical experiments, we demonstrated that the proposed arrangement facilitates accurate prediction regardless of the number of active elements.
- Open Access
- 28 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Error-Correcting Codes in the Two-Dimensional Lee Metric on Square and Hexagonal Signal Constellations
We construct linear codes over odd prime fields for correcting two-dimensional Lee-metric errors on the square and hexagonal signal constellations. They are obtained by puncturing and enlarging either Reed-Solomon codes or Bose-Chaudhuri-Hocquenghem codes. We introduce two-dimensional Lee weight on the square and hexagonal constellations in a similar way as the method presented by the first author in 2019 International Symposium on Information Theory, and propose an effective method for correcting two-dimensional Lee-metric errors of weight t>=1. The concept of value-locator of an error, which was introduced implicitly by K. Nakamura in 1979 and inherited to the above conference paper, is a key for decoding error-correcting Lee-metric codes. Our decoding method is based on the Buchberger algorithm for finding Groebner bases of ideals in the multivariate polynomial ring. The results of simulations show that our codes are more powerful for correcting two-dimensional Lee-metric errors in the two-dimensional signal constellations than the conventional codes for correcting Hamming-metric errors and one-dimensional Lee-metric errors, and the decoding method works well for correcting two-dimensional Lee-metric errors of small weight t.
- 28 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessError-Correcting Codes in the Two-Dimensional Lee Metric on Square and Hexagonal Signal Constellations
We construct linear codes over odd prime fields for correcting two-dimensional Lee-metric errors on the square and hexagonal signal constellations. They are obtained by puncturing and enlarging either Reed-Solomon codes or Bose-Chaudhuri-Hocquenghem codes. We introduce two-dimensional Lee weight on the square and hexagonal constellations in a similar way as the method presented by the first author in 2019 International Symposium on Information Theory, and propose an effective method for correcting two-dimensional Lee-metric errors of weight t>=1. The concept of value-locator of an error, which was introduced implicitly by K. Nakamura in 1979 and inherited to the above conference paper, is a key for decoding error-correcting Lee-metric codes. Our decoding method is based on the Buchberger algorithm for finding Groebner bases of ideals in the multivariate polynomial ring. The results of simulations show that our codes are more powerful for correcting two-dimensional Lee-metric errors in the two-dimensional signal constellations than the conventional codes for correcting Hamming-metric errors and one-dimensional Lee-metric errors, and the decoding method works well for correcting two-dimensional Lee-metric errors of small weight t.