招待講演1
岡部寿男（京都大学）
「戦略目標「Society 5.0時代の安心・安全・信頼を支える基盤ソフトウェア技術」の理念とJST CREST「S5基盤ソフト」の取り組み」
概要
文部科学省が定める戦略的創造研究推進事業の戦略目標の一つとして「Society 5.0時代の安心・安全・信頼を支える基盤ソフトウェア技術」が採択され、それに基づいて令和3年度からJST CREST「基礎理論とシステム基盤技術の融合によるSociety 5.0のための基盤ソフトウェアの創出」（S5基盤ソフト）とJSTさきがけ「社会変革に向けたICT基盤強化」（ICT基盤強化）の二つのプログラムが実施されている。また令和6年度からは同目標を含む7つの戦略目標に基づく新たなプログラムとしてJST ACT-X「AI共生社会を拓くサイバーインフラストラクチャ」（サイバーインフラ）が開始された。講演者は、JST CREST「S5基盤ソフト」の研究総括の立場から、戦略目標制定の経緯とCRESTでの取り組み、さきがけ「ICT基盤強化」との連携、NSF SaTC (Secure and Trustworthy Cyberspace)の動向などについて概観する。また、我が国におけるシステム基盤ソフトウェア研究の振興について、参加者と議論を深めたいと考えている。
招待講演2
小山田昌史（日本電気株式会社）
「LLMからAI Agentへ 〜RAG、自己進化、ワークフロー〜」
概要
大規模言語モデル（LLM）がチャットAIの形で世に広まり２年が経とうとしている。
講演者はNECにおいて独自の大規模言語モデルcotomiの研究開発リードを務め、「業務で使えるLLM」の方針のもと、グローバルでも競争力のある性能と速度を実現してきた。
他方、「業務で使える」という観点にもとづけばLLMのみでは不十分であり、外部からの知識の獲得、継続的なモデルの改善、外界との接続など、様々な発展がのぞまれる。
そのような背景から講演者の研究グループでは「ユーザからの要求にもとづき様々なシステムと連携しながら依頼をこなす自律的なAI Agent」の研究開発に取り組んでいる。
本講演ではLLMの開発からAI Agentに至るまでの流れと、AI Agentにおける先端の取り組みについてご紹介する。
トップカンファレンス凱旋講演
慶應義塾大学 土岐紘大 様
WoundDie: Concurrency Control Protocol with Lightweight Priority Control
Kodai Doki, Takashi Hoshino, and Hideyuki Kawashima,
In Proceedings of the 15th ACM SIGOPS Asia-Pacific Workshop on Systems (APSys ‘24), pp. 130–135 (2024).
https://doi.org/10.1145/3678015.3680480
Abstract
Recent concurrency control protocols for database transactions employ priority control to deliver predictable latency. Conventional protocols either require complex data structures or provide incomplete priority control that has a parameter and does not ensure starvation-freedom. To address this issue, we propose a new protocol, WoundDie, which ensures starvation-freedom without parameters. Unlike conventional starvation-free protocols, it does not require complex data structures for locking, making it memory-efficient and easy to implement. Experimental results demonstrated that under a high contention workload, WoundDie exhibited tail latency that was 6.5 times lower than Polaris’s best case and 80 times lower than its worst case.
東京農工大学 山田浩史 様
Reboot-Based Recovery of Unikernels at the Component Level
Takeru Wada and Hiroshi Yamada,
In 2024 54th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), pp. 15-28 (2024).
https://doi.org/10.1109/DSN58291.2024.00017
Abstract
The unikernel is a library operating system (OS) where OS functions are linked to the target applications. Making the unikernel layer as reliable as possible is mandatory because it controls the linked application’s execution. However, like commodity OS kernels, the unikernel suffers from software bugs and non-deterministic hardware failures. The current standard for recovering a failed unikernel is reboot-based and it involves restarting the whole unikernel-linked application, which leads to service stops and long downtimes. This paper presents VampOS that performs efficient reboot-based recovery of the unikernel layer. VampOS forces the unikernel components to interact with each other in a message-passing manner to restart only the damaged one while keeping the others and the application running. We prototyped VampOS on Unikraft 0.8.0 and QEMU 6.1.50. The experimental results show that the prototypes for four applications effectively recover the failed components with almost zero downtime.
慶應義塾大学 赤間滉星 様
Scrappy: SeCure Rate Assuring Protocol with PrivacY
Kosei Akama, Yoshimichi Nakatsuka, Masaaki Sato, Keisuke Uehara,
Network and Distributed System Security（NDSS）(2024).
https://doi.org/10.14722/ndss.2024.24445
Abstract
Preventing abusive activities caused by adversaries accessing online services at a rate exceeding that expected by websites has become an ever-increasing problem. CAPTCHAs and SMS authentication are widely used to provide a solution by implementing rate limiting, although they are becoming less effective, and some are considered privacy-invasive. In light of this, many studies have proposed better rate-limiting systems that protect the privacy of legitimate users while blocking malicious actors. However, they suffer from one or more shortcomings: (1) assume trust in the underlying hardware and (2) are vulnerable to side-channel attacks.
Motivated by the aforementioned issues, this paper proposes Scrappy: SeCure Rate Assuring Protocol with PrivacY. Scrappy allows clients to generate unforgeable yet unlinkable rate-assuring proofs, which provides the server with cryptographic guarantees that the client is not misbehaving. We design Scrappy using a combination of DAA and hardware security devices. Scrappy is implemented over three types of devices, including one that can immediately be deployed in the real world. Our baseline evaluation shows that the end-to-end latency of Scrappy is minimal, taking only 0.32 seconds, and uses only 679 bytes of bandwidth when transferring necessary data. We also conduct an extensive security evaluation, showing that the rate-limiting capability of Scrappy is unaffected even if the hardware security device is compromised.
慶應義塾大学 鈴木慶汰 様
Balancing Analysis Time and Bug Detection: Daily Development-friendly Bug Detection in Linux
Keita Suzuki, Kenta Ishiguro, Kenji Kono,
In 2024 USENIX Annual Technical Conference (ATC), pp. 493-508 (2024).
https://www.usenix.org/conference/atc24/presentation/suzuki
Abstract
Linux, a battle-tested codebase, is known to suffer from many bugs despite its extensive testing mechanisms. While many of these bugs require domain-specific knowledge for detection, a significant portion matches well-known bug patterns. Even though these bugs can be found with existing tools, our simple check of Linux kernel patches suggests that these tools are not used much in the developer’s daily workflow. The lack of usage is probably due to the well-known trade-off between analysis time and bug detection capabilities: tools typically employ complex analysis to effectively and comprehensively find bugs in return for a long analysis time, or focus on a short analysis time by only employing elementary analyses and thus can only find a very limited number of bugs. Ideally, developers expect the tools to incur short analysis time, while still finding many bugs to use them in daily development.
This paper explores an approach that balances this trade-off by focusing on bugs that can be found with less computationally-complex analysis methods, and limiting the scope to each source code. To achieve this, we propose a combination of computationally lightweight analyses and demonstrate our claim by designing FiTx, a framework for generating daily development-friendly bug checkers that focus on well-known patterns. Despite its simplicity, FiTx successfully identified 47 new bugs in the Linux kernel version 5.15 within 2.5 hours, outperforming Clang Static Analyzer and CppCheck in both speed and bug detection. It demonstrates that focusing on less complex bug patterns can still significantly contribute to the improvement of codebase health. FiTx can be embedded into the daily development routine, enabling early bug detection without sacrificing developers’ time.
