Self-Adapting Tool for Automated Receiver Test based on Artificial Intelligence
The aim of the START project is to develop an intelligent tool for automated conformance testing of digital multi-gigabit data transmission (>10 Gbit/s) on the PHY layer of receivers with a test signal that automatically adapts to the capabilities of a Device Under Test (DUT). For this purpose, an intelligent, self-learning algorithm for generating test patterns is being developed, which should maximize the bit error ratio (BER) to over 10^-6. The signal quality is also to be adjusted automatically and depending on the DUT by introducing specific impairments (e.g. jitter, noise). By evaluating the read error positions, it should be possible to identify weak points in the chip design of the receiver. The novel tool is to be further developed using an artificial neural network (ANN) and machine learning methods such as reinforcement learning, in order to continuously improve the search space for the test signals and patterns. This tool should help to avoid errors in chip design and accelerate design cycles. The project is being carried out in collaboration between IDACUS and BitifEye Digital Test Solutions GmbH.
