1. Introduction

HTG is a Hybrid systems Test Generation tool. The main functionality of the tool is to generate test cases in form of a tree from an input model.
The theoretical foundations of the tool can be summarized as follows: 
  • Formal framework for conformance testing of hybrid systems. This framework uses the commonly-accepted hybrid automaton model and allows, on one hand, to formally reason about the relation between a system under test and a specification, and on the other hand, to develop test generation algorithms.
  • Novel test coverage measure. This is a challenging problem in testing. Intuitively, test coverage is a way to characterize the relation between the number and the type of tests to execute and the portion of the system's behavior effectively tested. The tool HTG uses a novel coverage measure based on the star discrepancy from statistics (that measures the equidistribution degree of a set of states over the state space). 
  • Coverage-guided test generation. The tool includes an implementation of the test generation algorithm gRRT which is based on the RRT (Rapidly-exploring Random Tree) algorithm, a probabilistic motion planning technique in robotics. The gRRT algorithm includes a procedure for guiding the test generation process using the star discrepancy coverage measure.
  • Hybrid automata and SPICE circuits. Besides hybrid automata, the tool can accept SPICE netlists as input models. Furthermore, with view to applications in analog and mixed-signal circuits, an efficient and reliable simulation method is key. The state-of-the-art SPICE simulator is prone to convergence problems when dealing with circuit components with stiff characteristics. HTG is also connected to the platform SICONOS, developed at INRIA Grenoble, that contains a number of simulation algorithms based on the non-smooth approach, which have proved to be efficient for systems with stiff dynamics.
Applications. Besides traditional applications of hybrid systems, we explore a new domain which is analog and mixed signal circuits. Indeed, hybrid systems provide a mathematical model appropriate for the modeling and analysis of these circuits. The choice of this application domain is motivated by the need in automatic tools to facilitate the design of these circuits which, for various reasons, is still lagging behind the digital circuit design. Besides hybrid automata described using a textual language, the tool can accept as input electrical circuits specified using SPICE netlists. We have treated a number of case studies from control applications as well as from analog and mixed signal circuits. The experimental results obtained using the tool HTG show its applicability to systems with complex dynamics and its scalability to high dimensional systems (with up to 200 continuous variables).

References. See page Publications