Control system simulation using dsee high level instrument interface and behavioural description

Development of Karoo Array Telescope Control Protocol (KATCP) based control systems for the KAT-7 and MeerKAT radio telescopes proved the value of a fully simulated telescope system. Control interface simulators of all telescope subsystems were developed or sourced from the subsystems. SKA SA created libraries to ease creation of simulated KATCP devices. The planned SKA radio telescope chose the TANGO controls framework. To benefit from simulation-driven development tango-simlib, an OSS Python library for data-driven development of TANGO device simulators, is presented. Interface simulation with attributes only requires a POGO XMI file; more complex behaviour requires a simple JSON SIMDD (Simulator Description Datafile). Arbitrary behaviour is implemented selectively using Python code. A simulation-control interface for back-channel manipulation of the simulator for e.g. failure conditions is also generated. For the SKA Telescope Manager system an Eclipse DSEE (Domain Specific Engineering Environment) capturing the behaviour and interfaces of all telescope subsystems is being developed. The DSEE produces tango-simlib SIMDD files, ensuring that the generated simulators match their formal specification. INTRODUCTION The current MeerKAT Control And Monitoring (CAM) system is developed against a fully simulated telescope system. In development environments, all the subsystems that would make up the real telescope are simulated at the level of their KATCP interfaces. KATCP is a communications protocol based on top of the TCP/IP (Transfer Control Protocol/Internet Protocol) layer. It is a syntax specification for controlling devices over a TCP link. The full, actual, MeerKAT CAM is run against the simulated devices, thus CAM functionality to be tested without the need of real telescope hardware. This is exploited by CAM developers in their own development environments and also allows automated functional integration tests to be run daily. The testing of SKA Telescope Manager (TM) will be started in the absence of other Elements since not all of them will be available when TM is ready to be tested, it would be beneficial to have a mechanism that allows TM testing without depending on other element’s Local Monitoring and Controls (LMCs). A useful tool for developing an evolutionary TM prototype is a data-driven TANGO simulation framework that is used to develop Element LMC ∗ Work supported by NRF. † aramaila@ska.ac.za simulators that can be used in the TM test environment. The goal is to develop a simulation framework for Element LMC Simulators that can be used in the TM test environment. Furthermore, the TM interface to LMCs can itself can be simulated using this framework. The simulation framework was presented to Element Consortia to keep them abreast of developments in this respect. Element Consortia are being encouraged to make use of the Simulation Framework to develop LMC Simulators where required. The following risk reductions have been identified: • Risk reduction for early Assembly Integration and Verification (AIV) support; • Risk reduction for TM product development, by providing early LMC simulators; • Risk reduction for Element development by producing LMC Simulator Framework to aid them in the development of element simulators; • Risk reduction by producing an early scriptable TM Simulator that can aid Element LMC development and integration efforts. The early AIV integration would demand complete development of some components. Unavailability of hardware or incomplete development of element can be a hurdle for such AIV integration. The simulation framework reduces such risk by generating simulators that could be used in place of LMC’s which are not fully developed or unavailable due to hardware dependencies. The test framework provides an approach to create test cases and points out areas where it can reduce manual effort in writing test cases through some amount of automation. It was shown that a basic LMC simulator can be produced using the information provided by the Element LMC Interface Control Document (ICD) through the simulation data-description file. The approach also enhanced our understanding of how domain specific simulators can be integrated into the testing and simulation framework as and when they become available. The simulation framework demonstrates how this risk can be mitigated for the TM product development by auto generating the simulators for the LMC’s to a great extent based on the Self Description-Data (SDD) data that captures the information typically captured using ICD’s in a structured and machine processable manner. Initially this was an exploratory prototype with the aim to develop it further into an evolutionary prototype during 2017 in the period towards Critical Design Review [1]. 16th Int. Conf. on Accelerator and Large Experimental Control Systems ICALEPCS2017, Barcelona, Spain JACoW Publishing ISBN: 978-3-95450-193-9 doi:10.18429/JACoW-ICALEPCS2017-TUDPL03 TUDPL03 292 Co nt en tf ro m th is w or k m ay be us ed un de rt he te rm so ft he CC BY 3. 0 lic en ce (© 20 17 ). A ny di str ib ut io n of th is w or k m us tm ai nt ai n at tri bu tio n to th e au th or (s ), tit le of th e w or k, pu bl ish er ,a nd D O I. Software Technology Evolution APPROACH AND STRATEGY As TANGO has been selected as the LMC common framework this prototype can be developed using TANGO while incorporating concepts from the MeerKAT fully simulated framework as used in the MeerKAT CAM development and qualification. The usage of simulators during the development of MeerKAT telescope proved useful. Hence the aim was to enable the reuse of the same simulators for TANGO based environment. Based on this idea, a generic TANGO simulator open source library [2] was implemented in the Python language. The simulation library (simlib) eases the development of simulators exposing a desired TANGO interface. The initial step explored the technological feasibility of implementing a development environment which is aligned with the proposed Telescope Management (TELMGT) design. Using a Domain Specific Language (DSL) to generate the SDD was explored by the TCS-R&I team showcasing the use of aDSL developed using the Model Driven Engineering (MDE) methodology, implemented as an Eclipse Plugin. The DSL enabled capturing SDD information that was proposed in the TELMGT design. The first version of the SDD template to capture the self-description data was released as a part of LMC Interface Guideline (LIG) document by the TELMGT team. In this phase the DSL was used and compared against the SKA schema to see if there was any problem generating an instance of the template using the DSL. The prototyping plan identified the following iterations for the LMC Interface Simulator Framework prototype: • Iteration 1 First demonstration with MeerKAT framework using Tango simulator(s) and TANGO -> KATCP translators; • Iteration 2 Evaluation of TANGO tool capabilities in the context of a MeerKAT-like radio telescope; • Iteration 3 Improved Simulators, including behaviour extension; • Iteration 4 Demonstration of the exploratory LMC Simulation. The main focus of the initial iteration was on: training and familiarisation with the TANGO framework; implementing simple TANGO based simulators (i.e. simulated "devices" exposing TANGO interfaces; and investigating how TANGO devices may be incorporated in the existing, fully functional KATCP based MeerKAT Control and Monitoring (CAM) environment. The MeerKAT CAM performs a function equivalent to the SKA Telescope Manager for MeerKAT. The main focus of the second iteration was on investigating Tango tool capabilities in the context of a real telescope (MeerKAT) TM system; progress towards data-driven TANGO device simulators with richer behaviour; making simlib more dynamic; investigating the Eclipse based Domain Specific Engineering Environment prototype (DSEE) as a source of data for data-driven simulators; and translating between KATCP and TANGO devices and clients. The primary use case was to allow the use/evaluation of TANGO tools alongside the existing KATCP based MeerKAT CAM. Another future application could be interfacing KATCP devices (potentially MeerKAT subsystems) to a TANGO based TM. TOOLS Since we were a distributed collaboration team based out of South Africa and India various tools were used to facilitate the development of the Control System Simulation Framework, such JIRA© for issue tracking and project management and Google Drive© and WEBEX web conferencing for association purposes. GitHub© is mainly used as version control of project code repository. PROJECT OUTPUT First Demonstration with MeerKAT Framework using Tango Simulators During the initial stage, simple TANGO based simulators were implemented, and we investigated how those device simulator can actually be incorporated in the existing, fully functional KATCP based MeerKAT CAM environment. As a proof of concept, a weather device simulator was implemented, i.e., Fig. 1. This simulator was configured to match the functionality of the existing MeerKAT weather station. Each TANGO attribute of the simulated weather station is backed by a simlib simulated quantity. Only the data parameters need to be specified; simlib automatically creates a simulated TANGO attribute that varies according to the specified data. During phase 1, slew-limited Gaussian random variable quantities and constant quantities are supported. Simulation parameters (e.g. min/max/mean) can be specified, as can the metadata (e.g. TANGO attribute label, description text, unit) that is used to configure the TANGO attribute backed by the simulated quantity. Each simulator created with simlib creates two TANGO interfaces. The simulated device interface [Fig. 1] mimics the TANGO interface that a real device would have presented to the TM. The simulation control interface [Fig. 2] is used to control the behaviour of the simulator. It can be used to change