DFTS2: Simulating Deep Feature Transmission Over Packet Loss Channels
This work addresses the need for reliable simulation tools in edge-cloud AI systems to develop error control strategies, though it is incremental as it builds on existing simulation and concealment methods.
The paper tackles the problem of simulating deep feature transmission over unreliable channels in edge-cloud collaborative intelligence by introducing DFTS2, a TensorFlow 2-based framework, and uses it to conduct the most comprehensive study of packet loss concealment methods for collaborative image classification models.
In edge-cloud collaborative intelligence (CI), an unreliable transmission channel exists in the information path of the AI model performing the inference. It is important to be able to simulate the performance of the CI system across an imperfect channel in order to understand system behavior and develop appropriate error control strategies. In this paper we present a simulation framework called DFTS2, which enables researchers to define the components of the CI system in TensorFlow~2, select a packet-based channel model with various parameters, and simulate system behavior under various channel conditions and error/loss control strategies. Using DFTS2, we also present the most comprehensive study to date of the packet loss concealment methods for collaborative image classification models.