Gianluca Chiozzi

The ALMA common software: a developer-friendly CORBA-based framework

The ALMA Common Software (ACS) is a set of application frameworks built on top of CORBA. It provides a common software infrastructure to all partners in the ALMA collaboration. The usage of ACS extends from high-level applications such as the Observation Preparation Tool [7] that will run on the desk of astronomers, down to the Control Software [6] domain. The purpose of ACS is twofold: from a system perspective, it provides the implementation of a coherent set of design patterns and services that will make the whole ALMA software [1] uniform and maintainable; from the perspective of an ALMA developer, it provides a friendly programming environment in which the complexity of the CORBA middleware and other libraries is hidden and coding is drastically reduced. The evolution of ACS is driven by a long term development plan, however on the 6-months release cycle the plan is adjusted based on incoming requests from ALMA subsystem development teams. ACS was presented at SPIE 2002[2]. In the two years since then, the core services provided by ACS have been extended, while the coverage of the application framework has been increased to satisfy the needs of high-level and data flow applications. ACS is available under the LGPL public license. The patterns implemented and the services provided can be of use also outside the astronomical community; several projects have already shown their interest in ACS. This paper presents the status of ACS and the progress over the last two years. Emphasis is placed on showing how requests from ACS users have driven the selection of new features.

Application development using the ALMA common software

The ALMA Common Software (ACS) provides the software infrastructure used by ALMA and by several other telescope projects, thanks also to the choice of adopting the LGPL public license. ACS is a set of application frameworks providing the basic services needed for object oriented distributed computing. Among these are transparent remote object invocation, object deployment and location based on a container/component model, distributed error, alarm handling, logging and events. ACS is based on CORBA and built on top of free CORBA implementations. Free software is extensively used wherever possible. The general architecture of ACS was presented at SPIE 2002. ACS has been under development for 6 years and it is midway through its development life. Many applications have been written using ACS; the ALMA test facility, APEX and other telescopes are running systems based on ACS. This is therefore a good time to look back and see what have been until now the strong and the weak points of ACS in terms of architecture and implementation. In this perspective, it is very important to analyze the applications based on ACS, the feedback received by the users and the impact that this feedback has had on the development of ACS itself, by favoring the development of some features with respect to others. The purpose of this paper is to describe the results of this analysis and discuss what we would like to do in order to extend and improve ACS in the coming years, in particular to make application development easier and more efficient.

Evolution of the VLT instrument control system toward industry standards

The VLT control system is a large distributed system consisting of Linux Workstations providing the high level coordination and interfaces to the users, and VME-based Local Control Units (LCUs) running the VxWorks real-time operating system with commercial and proprietary boards acting as the interface to the instrument functions. After more than 10 years of VLT operations, some of the applied technologies used by the astronomical instruments are being discontinued making it difficult to find adequate hardware for future projects. In order to deal with this obsolescence, the VLT Instrumentation Framework is being extended to adopt well established Commercial Off The Shelf (COTS) components connected through industry standard fieldbuses. This ensures a flexible state of the art hardware configuration for the next generation VLT instruments allowing the access to instrument devices via more compact and simpler control units like PC-based Programmable Logical Controllers (PLCs). It also makes it possible to control devices directly from the Instrument Workstation through a normal Ethernet connection. This paper outlines the requirements that motivated this work, as well as the architecture and the design of the framework extension. In addition, it describes the preliminary results on a use case which is a VLTI visitor instrument used as a pilot project to validate the concepts and the suitability of some COTS products like a PC-based PLCs, EtherCAT8 and OPC UA6 as solutions for instrument control.

Enabling technologies and constraints for software sharing in large astronomy projects

The new observatories currently being built, upgraded or designed represent a big step up in terms of complexity (laser guide star, adaptive optics, 30/40m class telescopes) with respect to the previous generation of ground-based telescopes. Moreover, the high cost of observing time imposes challenging requirements on system reliability and observing efficiency as well as challenging constraints in implementing major upgrades to operational observatories. Many of the basic issues are common to most of the new projects, while each project also brings an additional set of very specific challenges, imposed by the unique characteristics and scientific objectives of each telescope. Finding ways to share the solution and the risk for these common problems would allow the teams in the different projects to concentrate more resources on the specific challenges, while at the same time realizing more reliable and cost efficient systems. In this paper we analyze the many dimensions that might be involved in sharing and re-using observatory software (e.g. components, design, infrastructure frameworks, applications, toolkits, etc.). We also examine observatory experiences and technology trends. This work is the continuation of an effort started in the middle of 2007 to analyze the trends in software for the control systems of large astronomy projects.

Introducing high performance distributed logging service for ACS

The ALMA Common Software (ACS) is a software framework that provides the infrastructure for the Atacama Large Millimeter Array and other projects. ACS, based on CORBA, offers basic services and common design patterns for distributed software. Every properly built system needs to be able to log status and error information. Logging in a single computer scenario can be as easy as using fprintf statements. However, in a distributed system, it must provide a way to centralize all logging data in a single place without overloading the network nor complicating the applications. ACS provides a complete logging service infrastructure in which every log has an associated priority and timestamp, allowing filtering at different levels of the system (application, service and clients). Currently the ACS logging service uses an implementation of the CORBA Telecom Log Service in a customized way, using only a minimal subset of the features provided by the standard. The most relevant feature used by ACS is the ability to treat the logs as event data that gets distributed over the network in a publisher-subscriber paradigm. For this purpose the CORBA Notification Service, which is resource intensive, is used. On the other hand, the Data Distribution Service (DDS) provides an alternative standard for publisher-subscriber communication for real-time systems, offering better performance and featuring decentralized message processing. The current document describes how the new high performance logging service of ACS has been modeled and developed using DDS, replacing the Telecom Log Service. Benefits and drawbacks are analyzed. A benchmark is presented comparing the differences between the implementations.

Model based systems engineering for astronomical projects

Model Based Systems Engineering (MBSE) is an emerging field of systems engineering for which the System Modeling Language (SysML) is a key enabler for descriptive, prescriptive and predictive models. This paper surveys some of the capabilities, expectations and peculiarities of tools-assisted MBSE experienced in real-life astronomical projects. The examples range in depth and scope across a wide spectrum of applications (for example documentation, requirements, analysis, trade studies) and purposes (addressing a particular development need, or accompanying a project throughout many - if not all - its lifecycle phases, fostering reuse and minimizing ambiguity). From the beginnings of the Active Phasing Experiment, through VLT instrumentation, VLTI infrastructure, Telescope Control System for the E-ELT, until Wavefront Control for the E-ELT, we show how stepwise refinements of tools, processes and methods have provided tangible benefits to customary system engineering activities like requirement flow-down, design trade studies, interfaces definition, and validation, by means of a variety of approaches (like Model Checking, Simulation, Model Transformation) and methodologies (like OOSEM, State Analysis)

PC based PLCs and ethernet based fieldbus: the new standard platform for future VLT instrument control

ESO is currently in the final phase of the standardization process for PC-based Programmable Logical Controllers (PLCs) as the new platform for the development of control systems for future VLT/VLTI instruments. The standard solution used until now consists of a Local Control Unit (LCU), a VME-based system having a CPU and commercial and proprietary boards. This system includes several layers of software and many thousands of lines of code developed and maintained in house. LCUs have been used for several years as the interface to control instrument functions but now are being replaced by commercial off-the-shelf (COTS) systems based on BECKHOFF Embedded PCs and the EtherCAT fieldbus. ESO is working on the completion of the software framework that enables a seamless integration into the VLT control system in order to be ready to support upcoming instruments like ESPRESSO and ERIS, that will be the first fully VLT compliant instruments using the new standard. The technology evaluation and standardization process has been a long and combined effort of various engineering disciplines like electronics, control and software, working together to define a solution that meets the requirements and minimizes the impact on the observatory operations and maintenance. This paper presents the challenges of the standardization process and the steps involved in such a change. It provides a technical overview of how industrial standards like EtherCAT, OPC-UA, PLCOpen MC and TwinCAT can be used to replace LCU features in various areas like software engineering and programming languages, motion control, time synchronization and astronomical tracking.

A code generation framework for the ALMA common software

Code generation helps in smoothing the learning curve of a complex application framework and in reducing the number of Lines Of Code (LOC) that a developer needs to craft. The ALMA Common Software (ACS) has adopted code generation in specific areas, but we are now exploiting the more comprehensive approach of Model Driven code generation to transform directly an UML Model into a full implementation in the ACS framework. This approach makes it easier for newcomers to grasp the principles of the framework. Moreover, a lower handcrafted LOC reduces the error rate. Additional benefits achieved by model driven code generation are: software reuse, implicit application of design patterns and automatic tests generation. A model driven approach to design makes it also possible using the same model with different frameworks, by generating for different targets. The generation framework presented in this paper uses openArchitectureWare1 as the model to text translator. OpenArchitectureWare provides a powerful functional language that makes this easier to implement the correct mapping of data types, the main difficulty encountered in the translation process. The output is an ACS application readily usable by the developer, including the necessary deployment configuration, thus minimizing any configuration burden during testing. The specific application code is implemented by extending generated classes. Therefore, generated and manually crafted code are kept apart, simplifying the code generation process and aiding the developers by keeping a clean logical separation between the two. Our first results show that code generation improves dramatically the code productivity.

Bulk data transfer distributer: a high performance multicast model in ALMA ACS

A high performance multicast model for the bulk data transfer mechanism in the ALMA (Atacama Large Millimeter Array) Common Software (ACS) is presented. The ALMA astronomical interferometer will consist of at least 50 12-m antennas operating at millimeter wavelength. The whole software infrastructure for ALMA is based on ACS, which is a set of application frameworks built on top of CORBA. To cope with the very strong requirements for the amount of data that needs to be transported by the software communication channels of the ALMA subsystems (a typical output data rate expected from the Correlator is of the order of 64 MB per second) and with the potential CORBA bottleneck due to parameter marshalling/de-marshalling, usage of IIOP protocol, etc., a transfer mechanism based on the ACE/TAO CORBA Audio/Video (A/V) Streaming Service has been developed. The ACS Bulk Data Transfer architecture bypasses the CORBA protocol with an out-of-bound connection for the data streams (transmitting data directly in TCP or UDP format), using at the same time CORBA for handshaking and leveraging the benefits of ACS middleware. Such a mechanism has proven to be capable of high performances, of the order of 800 Mbits per second on a 1Gbit Ethernet network. Besides a point-to-point communication model, the ACS Bulk Data Transfer provides a multicast model. Since the TCP protocol does not support multicasting and all the data must be correctly delivered to all ALMA subsystems, a distributer mechanism has been developed. This paper focuses on the ACS Bulk Data Distributer, which mimics a multicast behaviour managing data dispatching to all receivers willing to get data from the same sender.

ACS sampling system: design, implementation, and performance evaluation

By means of ACS (ALMA Common Software) framework we designed and implemented a sampling system which allows sampling of every Characteristic Component Property with a specific, user-defined, sustained frequency limited only by the hardware. Collected data are sent to various clients (one or more Java plotting widgets, a dedicated GUI or a COTS application) using the ACS/CORBA Notification Channel. The data transport is optimized: samples are cached locally and sent in packets with a lower and user-defined frequency to keep network load under control. Simultaneous sampling of the Properties of different Components is also possible. Together with the design and implementation issues we present the performance of the sampling system evaluated on two different platforms: on a VME based system using VxWorks RTOS (currently adopted by ALMA) and on a PC/104+ embedded platform using Red Hat 9 Linux operating system. The PC/104+ solution offers, as an alternative, a low cost PC compatible hardware environment with free and open operating system.