发表机构
Chongqing University; National University of Defense Technology; Chongqing University of Posts and Telecommunications; University of Toronto Mississauga; Chongqing Huayuan Zhixin Technology Co., Ltd.; Inspur Yunzhou Industrial Internet Co., Ltd.; Guoqi Zhimo (Chongqing) Technology Co., Ltd.（重庆大学; 国防科技大学; 重庆邮电大学; 多伦多大学密西沙加分校; 重庆华远智信科技有限公司; 浪潮云洲工业互联网有限公司; 国器智模（重庆）科技有限公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
PACE框架通过联合控制级联路由和填充内容，最小化检索增强对话服务的感知首响应时间，在保持质量的同时显著降低延迟和成本。
AI 中文摘要
我们提出了PACE，一个用于检索增强对话服务的框架，该框架将感知首响应时间（PTFR）形式化为QoE目标，并在质量/成本约束下最小化该目标。与先前关于级联路由、语义缓存或自适应检索的工作不同，PACE联合控制哪个答案源组成响应以及什么内容填充等待窗口。该系统部署在一个人形机器人销售服务上，结合了三种机制：负载自适应级联路由器、联合路径填充控制器和波动感知缓存准入。在75k个CarQA请求上，级联在P95时将纯LLM的PTFR减半（在c16下为0.29秒对比0.53秒）。自适应控制器达到0.41秒的P95，在高负载下以相同质量比RAG快2.4倍。填充控制器将调用减少94%且零冲突。波动感知准入将过时答案从86%降至0%。一个门控规则确保控制器从不差于基线，且暴露时间限制在一个保持周期内。这是首次在部署服务中对填充答案冲突风险进行量化。
英文摘要
We present the PACE, a framework for retrieval-augmented dialogue serving that formalizes Perceived Time-to-First-Response (PTFR) as a QoE objective and minimizes it under quality/cost constraints. Unlike prior work on cascaded routing, semantic caching, or adaptive retrieval, PACE jointly controls which answer source composes the response and what fills the waiting window. Deployed on a humanoid-robot sales service, it combines three mechanisms: a load-adaptive cascading router, a joint path-filler controller, and volatility-aware cache admission. On 75k CarQA requests, the cascade halves pure-LLM PTFR at P95 (0.29 vs 0.53s at c16). The adaptive controller reaches 0.41s P95, outperforming RAG by 2.4 times at high load with equal quality. The filler controller cuts calls by 94% with zero conflict. Volatility-aware admission reduces stale answers from 86% to 0%. A gating rule ensures the controller never worse than the baseline, with exposure bounded by one hold period. This is the first quantification of filler-answer conflict risk in deployed services.
