Differential Fuzzing for Data Distribution Service Programs with Dynamic Configuration

This paper is an intermediate report on a progressive study of OMG DDS (Data Distribution Service), the data-centric publish-subscribe communication paradigm. The DDS has naturally distinct features such as asynchronous interaction, diverse QoS requirement and a dynamic discovery which are major factors of middleware infrastructure supporting mobile and pervasive applications. We propose a DDSS (Data Distribution and Storage Service) based on the DDS and persistent local data space with functional capability of dynamic reconfiguration. Data space is built on a main memory database system in order to provide high-performance and mobility support. DDSS aims at supporting applications enabling seamless service in heterogeneous, dynamically varying computing and communication environment.

In this paper, we present a service-oriented home network middleware based on open grid service architecture. It supports two services: automatic distribution and dynamic migration schemes which shall be exploited efficiently in ubiquitous home networking environment. The deployment of Open Grid Service Architecture (OGSA) allows our system to have several advantages: adaptability, dynamic configuration, transparency, fault tolerance and performance. Automatic distribution service supports adaptability, dynamic configuration and transparency by providing automatic allocation and execution of home network platforms. Dynamic migration service supports fault tolerance and performance by migrating the processes in the failed server or with poor performance.

In this paper we present an approach to timing analysis of applications communicating over a FlexRay bus, taking into consideration the specific aspects of the publish subscribe paradigm. We give an overview of works dealing with Data Distribution Service (DDS), scheduling over FlexRay and scheduling over the Control area Network (CAN). We are interesting in real time Quality of Service (QoS) for DDS. We specially focus on real time communication on static and dynamic FlexRay bus. We give our approach to calculate deadline on a DDS distribution using the equal Flexibility. We give also DDS and the task models for the computation. In Many distributed embedded real-time applications, like military, automobile, aeronautique data produced in one component of the system needs to be shared with other components of the system. The middleware Data Distribution Service DDS is divided into two components: the Data-Centric Publish-Subscribe DCPS layerand the Data local Reconstruction Layer DLRL. The DCPS layer transports data from publishers to subscribers according to Quality of Service QoS constraints associated with the data topic, publisher, and subscriber. The DLRL allows distributed data to be shared by local objects located remotely from each other as if the data is local. The DLRL is built on top of the DCPS layer. DDS impose of Service in transmission between publishers and subscribers. The transmission of data is managed by the real time bus. Real time applications use CAN or FlexRay bus. The challenge is to deal with a configuration using DDS middleware over FlexRay bus. The goal of using QoS in DDS is that all DataReaders read temporally valid data. DDS proveides QosPolicies specifically targeted to minimum latency, predictable real-time operation in high performance distributed data-critical systems. The DEADLINE QosPolicy expresses the maximum duration within which a DataReader expects a data-object instance to be updated. Our research team was interested in the LifeSpan (LS) Qos which concerns the Topic and the Data Writer entities. It specifies the maximum duration of validity of the data written by the DataWriter. They developed [6] an algorithm that calculate data distribution deadline. The following assumptions are considered: all nodes, data objects, and scheduling parameters are known a priori; each DataReader/DataWriter has a local node, where it originates.; the period of a DataWriter is always less than its Lifespan; and the offered deadline period is always less then the requested deadline period. This work gives an analysis for deadline computation on a FlexRay bus. The goal is to consider the data-centric publish- subscribe (DCPS) and to use DDS Qos. Both Dynamic segment and static segment are considered. We are currently working on a project to implement deadline computation with C language in order to finalize our works dealing with our DDS implementation.

The OMG Data Distribution Service (DDS) is an emerging specification for publish-subscribe data-distribution systems. The goal of DDS is to facilitate the efficient distribution of data. The integration of DDS allows also to take advantage of the rich set of Quality of Service (QoS) parameters and events generated in response to faults or a violation of temporal parameters. However, the level of abstraction they supply make it hard to configure DDS. For example System developers must manually manage subscriptions and must manually configure the publishers and subscribers. To reduce complexity associated with DDS configuration we focused on a new design based on the model Based Design. Indeed, we designed a new Simulink DDS Blockset and we integrated the required DDS entities in a vehicle application. In this paper, we present a new design of the Electronic Control Units (ECU) of modern vehicle based to an extended model of the Society of Automotive Engineers (SAE) benchmark. Second, we extended each vehicle module by the designed DDS blocks. Finally, we generate the xml description of DDS relative to our automotive application. To explain our approach, we detail in this paper the implementation of an Active Frame Steering Controller and the integration of corresponded DDS entities.

Ubiquitous or pervasive computing is a new kind of computing, where specialized elements of hardware and software will have such high level of deployment that their use will be fully integrated with the environment. Augmented reality extends reality with virtual elements but tries to place the computer in a relatively unobtrusive, assistive role. To our knowledge, there is no specialized network middleware solution for large-scale mobile and pervasive augmented reality games. We present here work that focus on the creation of such network middleware for mobile and pervasive entertainment, applied to the area of large scale augmented reality games. In, this context, mechanisms are being studied, proposed and evaluated to deal with issues such as scalability, multimedia data heterogeneity, data distribution and replication, consistency, security, geospatial location and orientation, mobility, quality of service, management of networks and services, discovery, ad-hoc networking and dynamic configuration

The Data Distribution Service (DDS) was started by the Object Management Group (OMG) in 2004. Currently, DDS is one of the contenders to support the Internet of Things (IoT) and the Industrial IOT (IIoT). DDS has also been used as a distributed simulation architecture. Given the anticipated proliferation of IoT and II devices, along with the explosive growth of sensor technology, can we expect this to have an impact on the broader community of distributed simulation? If it does, what is the impact and which distributed simulation domains will be most affected? DDS shares many of the same goals and characteristics of distributed simulation such as the need to support scale and an emphasis on Quality of Service (QoS) that can be tailored to meet the end user’s needs. In addition, DDS has some built-in features such as security that are not present in traditional distributed simulation protocols. If the IoT and II realize their potential application, we predict a large base of technology to be built around this distributed data paradigm, much of which could be directly beneficial to the distributed M&S community. In this paper we compare some of the perceived gaps and shortfalls of current distributed M&S technology to the emerging capabilities of DDS built around the IoT. Although some trial work has been conducted in this area, we propose a more focused examination of the potential of these new technologies and their applicability to current and future problems in distributed M&S. The Internet of Things (IoT) and its data communications mechanisms such as the Data Distribution System (DDS) share properties in common with distributed modeling and simulation (M&S) and its protocols such as the High Level Architecture (HLA) and the Test and Training Enabling Architecture (TENA). This paper proposes a framework based on the sensor use case for how the two communities of practice (CoP) can benefit from one another and achieve greater capability in practical distributed computing.

DDS (Data Distribution Service) is a set of API and interoperability protocol specifications developed by OMG. It adopts a datacentric publish/subscribe architecture to meet the demand of high efficiency and real-time communication, This paper designed a kind of DDS based on data distribution system and used in distributed simulation system at first through the analysis of DDS communication principle. Then, it achieved the system from the design of communication interface to the construction of distributed simulation system. And by means of Quality of Service (QoS) Strategy in DDS, the particularly modified QoS Strategy designed for distributed simulation system is proposed. Finally, through the analysis on effect of disturbance (system average transmission delay, jittering and the amount of packet loss) on real-time and reliability requirements of system, the test results indicates that The distributed simulation system based on DDS can meet the real-time and reliability of general distributed systems.

The Southern African Large Telescope (SALT) software control system1 is realised as a distributed control system, implemented predominantly in National Instruments' LabVIEW. The telescope control subsystems communicate using cyclic, state-based messages. Currently, transmitting a message is accomplished by performing an HTTP PUT request to a WebDAV directory on a centralised Apache web server, while receiving is based on polling the web server for new messages. While the method works, it presents a number of drawbacks; a scalable distributed communication solution with minimal overhead is a better fit for control systems. This paper describes our exploration of the Data Distribution Service (DDS). DDS is a formal standard specification, defined by the Object Management Group (OMG), that presents a data-centric publish-subscribe model for distributed application communication and integration. It provides an infrastructure for platform- independent many-to-many communication. A number of vendors provide implementations of the DDS standard; RTI, in particular, provides a DDS toolkit for LabVIEW. This toolkit has been evaluated against the needs of SALT, and a few deficiencies have been identified. We have developed our own implementation that interfaces LabVIEW to DDS in order to address our specific needs. Our LabVIEW DDS interface implementation is built against the RTI DDS Core component, provided by RTI under their Open Community Source licence. Our needs dictate that the interface implementation be platform independent. Since we have access to the RTI DDS Core source code, we are able to build the RTI DDS libraries for any of the platforms on which we require support. The communications functionality is based on UDP multicasting. Multicasting is an efficient communications mechanism with low overheads which avoids duplicated point-to-point transmission of data on a network where there are multiple recipients of the data. In the paper we present a performance evaluation of DDS against the current HTTP-based implementation as well as the historical DataSocket implementation. We conclude with a summary and describe future work.

Data Distribution Service (DDS) has emerged as a potential solution for data communication challenges in smart grids. DDS is designed to support quality communication for large scale real-time systems through a wide range of QoS policies. However, a smart grid involves various types of communication applications running on different computing environments. Some environments have limited computing resources such as small memory and low performance, which makes it difficult to accommodate DDS. In this paper, we present a feature-based approach for tailoring DDS to configure lightweight DDS by selecting only the necessary features for the application in consideration of the resource constraints of its running environment. This allows DDS to serve as a uniform communication middleware across the smart grid, which is critical for interoperability. We analyze DDS in terms of features and design them using Unified Modeling Language (UML) and Object Constraint Language (OCL) based on inheritance and overriding. We define a formal notion of feature composition to build DDS configurations. We implemented the approach in OpenDDS and demonstrate its application to different application environments. We also experimented the approach for the efficiency of configured DDS in terms of resource utilization. The results show that configured DDS is viable for efficient and quality data communication for applications that run on an environment with limited computing capability.

The data distribution service (DDS) standard enables heterogeneous applications to interact in real-time distributed systems with high-performance requirements. Real-time networking must provide predictable latencies in its messaging mechanisms, in order to allow participants to perform its schedulability analysis. However, there is a lack of support for scheduling DDS applications over real-time networks. This paper describes the design and implementation of a real-time data distribution strategy that uses the object management group DDS and real-time network facilities to ensure the on-time and temporally valid delivery of data. This strategy implements an algorithm, called the LifeSpan-based consistent data delivery algorithm to determine scheduling parameters that will ensure that data that arrives at a requesting target is valid. This implementation introduces a global scheduling solution that manages the distribution and transmission activities in order to ensure the respect of DDS entities deadlines.

Rapid growth of distributed services in heterogeneous environments calls for an adaptive and unified QoS framework, which allows flexibility and configurability in distributed service provisioning. In this paper, we present a resource-aware and component-based middleware called 2K Q. 2KQ enables active and configurable distributed services in highly dynamic and heterogeneous environments. The key idea of 2K Q is dynamic and resource-aware service configuration. Multiple service configurations are defined for the delivery of one distributed service: each service configuration consists of a different set of service components. At runtime, 2K Q dynamically select an appropriate service configuration for each service request, based on the current end-to-end resource availability condition. If a service component is not available on a host, 2K Q will initiate active downloading of this service component. The selected service configuration performs customized service delivery with satisfactory QoS to the requesting client. Our prototype of 2K Q as well as an example Video Streaming service on top of it demonstrate the soundness of 2K Q.Key wordsQoSmiddlewaredistributed serviceservice configurationactive service component downloading

International audience

Bloom filter-based discovery protocol for DDS middleware

The Southern African Large Telescope (SALT) telescope control system is realised as a distributed control system, implemented predominantly in National Instruments’ LabVIEW. The telescope subsystems communicate using cyclic, state-based messages. Previously, we explored Data Distribution Service (DDS)—a data-centric publish-subscribe model for distributed application communication and integration, with the aim of using it as our primary communications middleware; however, obstacles uncovered during the implementation and roll-out of DDS prompted us to take a fresh look at the problem. We identified ZeroMQ (ZMQ) and NATS as possible alternative solutions. While ZMQ and NATS are both high-performance asynchronous messaging systems similar to DDS, the main difference between them is that ZMQ, in resemblance to DDS, is brokerless, whereas NATS is brokered. In a brokered system, the message broker mediates communication between a sender and a receiver, minimising the mutual awareness that applications need to have of each other in order to be able to exchange messages. Furthermore, less complexity is required in both the sender and the receiver of a message, when making use of a broker. A drawback of a broker, however, is that it may introduce a possible single point of failure. As a result of the simplicity of the client, coupled with the succinct NATS wire protocol, we were able to incorporate the NATS sender and receiver logic natively in our LabVIEW software. Taking advantage of a native solution removes the maintenance and deployment burden introduced by the use of an external C-language library, which is required for the use of both DDS and ZMQ. During our investigation of ZMQ we uncovered the same obstacles as we did with DDS, which involved the complexities of gracefully shutting down the communications logic when not using an object-oriented programming methodology. Due to legacy software, we have to maintain backward compatibility with a version of LabVIEW which predates the availability of built-in object-orientation in the language. This, coupled with the advantages of a native solution, led to the adoption of NATS as our messaging system. In this paper, we explore ZMQ and NATS as alternatives to DDS, accompanied with results of a performance evaluation of DDS, ZMQ and NATS against our home-grown legacy HTTP-based messaging implementation. We conclude with a summary and describe future work.

Cyber-Physical Systems (CPS) represent next generation distributed and embedded systems. Robot Operating System (ROS), an open-source middleware for robotics development, has been widely used for CPS applications. However, ROS is not suitable for real-time and embedded systems, because ROS does not support to meet real-time requirements, and runs only on a few kinds of OSs. Facing this problem, ROS is going to be significantly upgraded as ROS2 with utilizing Data Distribution Service (DDS). DDS is suitable for CPS due to its various transport configurations (e.g. reliability and durability). ROS2 needs to convert data for DDS and abstracting DDS from ROS2 users, which causes the overhead examined in this research. Transport latency between ROS2 nodes varies by transport situations, data size, and DDS vendors. We clarify the performance characteristics of currently-available data transport on ROS or ROS2 in various situations. Revealing the present capability of ROS2 depending on DDS vendors and DDS configurations, we explore and evaluate the facing constraints and potential of ROS2.