CKN Edge AI Dataset for Image inference at the Edge (CEAD)

This synthetic workload models camera device requests for resource constrained inference requests at the Edge for Campaign Knowledge Network evaluation. The workload is a deterministic and pre-ordered set of time windows containing close to 5 million individual data points belonging to 1500 time windows, each time window with a number of requests between 100-1000. Composed of independent inference requests (events), the workload is structured to reflect sudden changes in need as reflected by the user-perceived quality of experience (e.g., accuracy and latency).本合成工作负载面向活动知识网络（Campaign Knowledge Network）的评估任务，模拟边缘场景下摄像设备针对资源受限推理请求所发起的请求行为。该工作负载为一组经预排序的确定性时间窗口集合，涵盖1500个时间窗口下的近500万个独立数据点，每个时间窗口内的请求数量介于100至1000之间。该工作负载由独立的推理请求（事件）构成，其结构设计旨在还原由用户感知体验质量（如准确率与延迟）所反映的需求突发波动。