| IEICE Technical Committee Submission System Conference Paper's Information |
Online Proceedings [Sign in] Tech. Rep. Archives |
| Go Top Page | Go Previous | [Japanese] / [English] |
| Paper Abstract and Keywords | ||
| Presentation | 2025-07-31 11:10 Traffic Analysis based on Periodicity Analysis using Graphs for Device Type Classification Chikako Takasaki, Tomohiro Korikawa, Takaaki Moriya, Kyota Hattori (NTT)  NS2025-60 |
|
| Abstract | (in Japanese) | (See Japanese page) |
| (in English) | In the beyond 5G and 6G networks, the number of connected devices and their types of services will greatly increase including not only user devices such as smartphones but also the Internet of Things (IoT). Therefore, the existing common control for all devices cannot fulfill the network requirements of each device; however, it is not feasible to control networks per device in terms of computational complexity. We introduce device types as groups of devices with similar traffic characteristics to control network per device type for efficient network control. We present a method to classify device types based on periodicity analysis of only encrypted traffic behavior. In real networks, feature engineering, such as manual feature selection and parameter optimization depending on input traffic, is not feasible because operators must treat a large amoount of traffic. This paper proposes a graph-based traffic classification method that trains only two-dimensional features: packet inter-arrival times and packet sizes. The proposed method extracts the typical length of the traffic, which is different depending on the input traffic, using periodicity analysis. The traffic for the extracted length is converted into graphs whose two adjacent nodes have an edge, which is defined as textit{linear graphs} in this paper. The inter-arrival times and the packet sizes are represented as node features and edge weights, respectively. The evaluation results show that the proposed method maintains 95-99% accuracy even trained with two-dimensional features, which is less than the numbers of features used in the existing methods, for two open datasets. Therefore, the proposed method is expected to reduce the feature engineering while maintaining the accuracy. |
|
| Keyword | (in Japanese) | (See Japanese page) |
| (in English) | device type classification / traffic analysis / periodicity analysis / graph neural network / / / / | |
| Reference Info. | IEICE Tech. Rep., vol. 125, no. 132, NS2025-60, pp. 40-45, July 2025. | |
| Paper # | NS2025-60 | |
| Date of Issue | 2025-07-23 (NS) | |
| ISSN | Online edition: ISSN 2432-6380 | |
| Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| Download PDF | NS2025-60 |
|
| Conference Information | |
| Committee | SR RCS RCC SeMI NS RISING HCL |
| Conference Date | 2025-07-30 - 2025-08-01 |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | MALIOS |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Paper Information | |
| Registration To | NS |
| Conference Code | 2025-07-SR-RCS-RCC-SeMI-NS-RISING-HCL |
| Language | Japanese |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Traffic Analysis based on Periodicity Analysis using Graphs for Device Type Classification |
| Sub Title (in English) | |
| Keyword(1) | device type classification |
| Keyword(2) | traffic analysis |
| Keyword(3) | periodicity analysis |
| Keyword(4) | graph neural network |
| Keyword(5) | |
| Keyword(6) | |
| Keyword(7) | |
| Keyword(8) | |
| 1st Author's Name | Chikako Takasaki |
| 1st Author's Affiliation | NTT (NTT) |
| 2nd Author's Name | Tomohiro Korikawa |
| 2nd Author's Affiliation | NTT (NTT) |
| 3rd Author's Name | Takaaki Moriya |
| 3rd Author's Affiliation | NTT (NTT) |
| 4th Author's Name | Kyota Hattori |
| 4th Author's Affiliation | NTT (NTT) |
| Speaker | Author-1 |
| Date Time | 2025-07-31 11:10:00 |
| Presentation Time | 25 minutes |
| Registration for | NS |
| Paper # | NS2025-60 |
| Volume (vol) | vol.125 |
| Number (no) | no.132 |
| Page | pp.40-45 |
| #Pages | 6 |
| Date of Issue | 2025-07-23 (NS) |
[Return to Top Page]