已知源标识符：分布式应用的三级身份系统
Source-Known Identifiers: A Three-Tier Identity System for Distributed Applications
AI总结：
本文提出SKIDs三级身份系统，解决分布式应用中存储效率、时间排序、元数据嵌入、零查找验证、保密性和多世纪地址性等六个需求。
摘要：
分布式应用需要满足存储效率、时间排序性、元数据嵌入、零查找验证性、外部消费者保密性和多世纪地址性的标识符。基于文献综述，现有方案无法在统一系统中提供所有六个属性。本文引入已知源标识符（SKIDs），一种三级身份系统，跨信任边界投影单一实体身份，解决所有六个属性。第一层，已知源标识符（SKID）是64位带符号整数，嵌入具有250毫秒精度的时间戳、应用拓扑和每实体类型序列计数器。它作为数据库主键，提供紧凑存储（8字节）和自然B树排序用于优化数据库索引。第二层，已知源实体标识符（SKEID）通过添加实体类型判别器、纪元选择器和BLAKE3 keyed消息认证码（MAC）将SKID扩展为128位通用唯一标识符（UUID）兼容值。SKEIDs在受信任环境中实现标识符起源、完整性和实体类型的零查找验证，具有大端字节布局，保持字典序UUID字符串比较中的时间顺序。第三层，安全SKEID通过AES-256对称加密加密整个SKEID，生成不可区分于随机字节的密文，同时保持标准UUID数据类型解析器的字符串表示兼容性。确定性的双向转换连接所有三层。
英文摘要：
Distributed applications use identifiers across database storage, trusted communication, and external access. These roles require different combinations of storage efficiency, chronological sortability, origin metadata embedding, zero-lookup verification, metadata confidentiality, and multi-century addressability. The identifier schemes compared in this paper, including Universally Unique Identifier (UUID) versions 4 and 7, do not individually provide all six properties. We present Source-Known Identifiers, a three-tier identity system that distributes these properties across representations of one entity identity, connected through deterministic transformations. A 64-bit Source-Known ID (SKID) provides a timestamp-ordered database key. A 128-bit Source-Known Entity ID (SKEID) adds entity type, epoch, and a keyed message authentication code (MAC) to the SKID, enabling integrity and origin metadata checks within a shared-key trust domain without record lookups. Secure SKEID encrypts this representation to conceal metadata while retaining linkability of repeated identifiers. Collision checks prevent generated ciphertext from being accepted as plaintext. Evaluation combines specification and threat analysis, implementation tests, and microbenchmarks of the C\#/.NET 10 reference implementation. Within sequence capacity on Apple M2 hardware, SKEID and Secure SKEID generation averaged 207.28 $\pm$ 3.10 ns and 224.62 $\pm$ 3.90 ns, respectively, compared with 361.94 $\pm$ 1.94 ns for UUID Version 4 and 384.11 $\pm$ 1.95 ns for UUID Version 7. $\pm$ values indicate 99.9% confidence interval margins. Under saturation, generation was slower than the UUID baselines. These results support the feasibility of deriving SKEIDs from 64-bit Source-Known IDs on demand.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
