SMARTDELTA

automated solutions for delta quality assessment

The SmartDelta project focuses on developing automated solutions for assessing the quality of product deltas within a continuous engineering environment. It provides intelligent analysis of development artefacts — such as source code, log files, or requirements specifications — and system execution data. This makes it possible to detect improvements or degradations in the quality of different product versions, enabling optimisation of subsequent builds. To achieve this, we have based the development of these new solutions on artificial intelligence, focusing on the verification and validation of quality characteristics in complex software development projects. In this way, we automate component reuse analysis through natural language processing techniques and the automated analysis of build recommendations using machine learning.

Challenges

Automatically addressing quality attributes continuously, from design through to operations.

Analysing, visualising, and transforming large volumes of data into actionable insights that can enhance the software quality of developed systems.

Identifying quality control measures that demonstrate key quality attributes without requiring a complete redesign of test cases.

Traditional reuse analysis approaches lack automated support for optimising quality characteristics.

Many traditional analysis and verification methods do not support continuous feedback loops.

Solution

SmartDelta combines rapid analysis and verification solutions through modelling and static analysis capabilities, test generation, selection, execution, and visualisation — enabling organisations to confidently deliver high‑quality systems in a fast‑paced agile environment.

SmartDelta aims to optimise analysis and verification activities across multiple product versions and deltas by automatically creating and refining verifiable models directly from development artefacts (requirements, logs, code, test records) and using these models to check non‑functional properties against system models. It also generates artefacts such as tests or monitors that can later be reused in the product line process for future variants.

We pursue the following technical objectives:

1. Create and extract models from different software artefacts, and verify consistency across deltas within a model‑chain architecture. This includes developing techniques for creating and extracting models from software artefacts based on usage scenarios and logs.

2. Implement automated consistency verification and validation based on model testing and model verification techniques. We provide novel approaches that enable automated verification of delta products using model specifications during both development and runtime.

3. Develop an automated, delta‑aware approach for reuse analysis of product line features, including trade‑off analysis between different non‑functional architectural properties of interest.

4. Enable automated test generation and analysis for non‑functional properties (e.g. resource consumption, security, performance) based on model specifications of the corresponding requirements, addressing real industrial challenges in continuous development.

5. Provide intelligent, adaptive, and ML‑based analysis, visualisation, and recommendations for upcoming delta versions of the system. Shorten development, deployment, and feedback cycles by applying quality control and trend analyses to deliver practical recommendations and identify vulnerabilities in system models early in the product lifecycle.

The effectiveness of SmartDelta is demonstrated through several complementary use cases, each targeting a set of typical scenarios envisioned for the proposed framework.