Real-Time Test Data Analysis for Test Optimization
TÜBİTAK grant for the project of Asst. Prof. Barış Arslan from ŞEHİR Computer Science and Engineering Department
Asst. Prof. Barış Arslan from Computer Science and Engineering Department, has been awarded a TÜBİTAK 3501 - Career Development Program grant for his project entitled "Real-Time Test Data Analysis for Test Optimization".

The design and productization of an Integrated Circuit (IC) in an intensely competitive semiconductor market requires extensive expertise in every phase (i.e., architecture, design, verification, manufacturing and testing) of the IC development process, not only delivering the best products but also doing it cost efficiently. The objective of this project is to develop data-driven test cost and quality optimization methods, particularly applicable to the chips that go into Internet of Things (IoT) and consumer electronics such as smartphones although it can be applied in the other sectors as well with appropriate test cost and quality trade-off points. The reason for this emphasis is that consumer electronics and IoT are highly competitive, low cost, low margin but high volume markets, driving the growth in semiconductor market. Low margins necessitate a tremendous vigilance at test cost but high volumes allow reaping the benefits of any test cost reduction for a large set of chips, magnifying the extend of cost saving.

Proposed project, with a novel approach, focuses on modelling the observable effects of defects on test effectiveness instead of trying to diagnose and model the defects themselves, moving test optimization problem to a higher-level of abstraction while heavily relying on test data. While pursuing this goal, it employs a novel interdisciplinary approach to test cost reduction by utilizing knowledge and techniques from real-time data stream analysis, machine learning, and IC testing.

It is expected that the research activities through the proposed project will make great strides in developing data-driven methods to address the challenging test cost issue with a high industrial relevance, particularly in high-volume consumer electronics and IoT.