Transportation Systems Applications
High-Reliability Communications Infrastructure with Tunable QoS
Real-time distributed connectivity will forever change transportation. From smart cars to intelligent tracking systems, things that move fast are quickly moving into the Internet of Things. Real-time communications will connect the components of cars together into intelligent machines. Those smart cars will integrate vehicles with traffic control into truly intelligent infrastructures. And, with real-time communications, companies will be able to keep track of their distributed fleets as never before.
Of course, the transportation industry is about more than just about moving vehicles. Manufacturing, test and training will also be transformed by easy connectivity.
This range of applications challenges the architecture. RTI Connext DDS rises to that challenge. It offers many capabilities needed by transportation systems. Connext DDS:
- Is flexible. Its pluggable transport and discovery adapt to the many types of networks needed by transportation systems. It supports 80+ platforms over 12 transports, transparently connecting most any systems.
- Works locally and remotely. It is equally at home in a tight vehicle feedback loop or connecting distant points over flaky channels. With facile routing, these systems combine into a single distributed system of systems, easing the transition from local control to global integration.
- Is truly reliable. The platform naturally supports redundant data sources and sinks. It arbitrates between conflicting publishers and fails over to backup sources instantly. It keeps multiple consumers in sync. It runs easily over redundant networks, intelligently delivering only one copy of received data.
- Supports huge fanout. With the only standardized reliable multicast protocol, Connext DDS can provide updates to thousands of endpoints efficiently.
- Is certifiable. Connext DDS Cert is a safety-certifiable communications infrastructure with an available DO-178C Level A Certification Package, that can be used as part of a flight-critical application in national air space..
Transportation systems are fundamentally distributed and real-time, so many transportation problems map well to the DDS real-time publish-subscribe pattern. Connext DDS is a proven substrate for vehicle integration and control, for inter-vehicle interactions, for tracking and control infrastructure. RTI Connext DDS is moving transportation into the Industrial Internet of Things.
Back in 2005, a Stanford team lead by Dr. Sebastian Thrun won the DARPA Grand Challenge by autonomously driving a modified Volkswagen Touareg named "Stanley" 175 miles in the Mojave Desert. Stanley is now housed in the Smithsonian museum, and Dr. Thrun is leading Google's driverless vehicle effort. But Stanley's software, originally funded by VW, lives on in Germany.
The autonomous vehicle algorithms are part of Volkswagen's effort in driver assistance and integrated safety. The system combines radars, laser range finders and video to assist safe operation. It helps avoid obstacles, detect lane departures, track eye activity and safely negotiate bends. RTI Connext DDS connects all the components into a single intelligent machine.
In 2013, VW demonstrated a version of this software running on an electric vehicle. It was able to drop off its passengers, drive autonomously down the block to a recharging station and then return fully charged. Driverless car technology is rapidly becoming reality.
To make this work, VW had to advance the state of the art in robotics research. To make it practical, VW also had to advance the state of the art of vehicle networking. Advanced technology requires fast Ethernet networks. However, automotive components use the CAN bus. This is a classic integration problem and one where RTI middleware shines. VW implemented a simple bridge running on a Linux-based ECU called CarGate. CarGate has an Ethernet port on one side and CAN networks on the other. It loads the popular ".dbc" CAN specification files and automatically generates DDS code to match. Thus, the fast networking components running DDS over Ethernet can easily subscribe to the left-rear-wheel speed and publish commands to the braking system, since both are connected to the CAN bus. RTI middleware naturally bridges high speed networking to the CAN bus.
Volkswagen's Driver Assistance and Integrated Safety system uses Connext DDS to combine radars, laser range finders, and video to assist safe operation. It tracks the driver's eyes, shown here, to detect drowsiness. It also detects lane departures, avoids collisions, and helps keep the car in its lane.
The VW CarGate ECU shuttles DDS messages seamlessly on and off the vehicle CAN bus.
The Volkswagen Group, the world's largest automotive company, emphasizes quality and advanced technology. To achieve those goals, Audi is building a new, state-of-the-art "hardware in the loop" simulator architecture. The simulator is fitted with electronic panels, each containing one Electronic Control Unit (ECU). A modern car may have 500 ECUs and over 100 processors.
Everything in the car with a wire is mounted on an ECU panel. These include engine control with fuel injectors and spark plugs, the driver console, Infotainment, antilock brakes, active suspension and more. Even the seat heaters are represented. The goal is to faithfully reproduce all the electrical systems in the vehicle and make sure they work together. The simulation must run fast enough to duplicate the real-time signals in an actual moving vehicle. When a test runs, each module can represented as either the actual hardware or a software simulation. If actual hardware fails during the simulation (a likely event, with 500 ECUs all under development), then the software simulation takes over so the test can continue.
The system is very flexible. Audi's system simulates whole cars, and even multiple cars interacting with each other (V2V) and the road infrastructure (V2I). ECUs can be exchanged to represent different vehicle options and different vehicles. The system will be used company-wide, on all of their brands, including Audi, Volkswagen, SEAT, Skoda, Lamborghini, Porsche, and Bentley.
Today their system uses a proprietary vendor architecture on a fiber network. For the future, Audi needs to combine multiple simulation vendors' systems into a single network. Connext DDS, with its blazing speed and easy integration technology, makes this possible.
A modern car contains hundreds of electronic modules and processors. Audi tests their systems under development with hardware-in-the-loop simulation. The simulation feed realistic data to actual components. RTI middleware enables a modular test environment that scales to work with hundreds of devices
Audi's test architecture integrates simulation test products from many manufacturers (herstellers). It implements two DDS domains: one for very fast simulation, and one to provide data to control the system and share data with storage systems and operators. The old architecture used a test rig server (prüfstandmanager) to shuttle data; with DDS, every device can just join the bus, so it's not needed.
For complete details, read A New Architecture for Automotive Hardware-in-the-Loop Test (courtesy ATZ elektronik).
NAV Canada, the air traffic control authority for Canada, is implementing continent-wide air traffic control with Connext DDS.
Canada is the world's second largest air navigation service provider (ANSP). The Canadian Automated Air Traffic System (CAATS) spans seven major centers and will connect to hundreds of airport towers. It controls from coast to coast, from the US border to the North Pole, and oceanic travel in the North Atlantic and Pacific. Performance, scalability and 24x7 reliability across this huge geography are key.
CAATS is a complex system that simultaneously enhances safety and improves service for customers. It provides tracking, collision detection and avoidance, and situational awareness for front-line professional controllers every day.
CAATS automates flight profile monitoring and extends conflict prediction and detection into non-radar airspace. It also processes and distributes flight data information to other NAV CANADA and international systems, enabling collaborative decision-making in flight planning.
CAATS also automatically updates flight information coming from other centers, computing flight estimates and processing flight plans.
The elimination of many manual processes ‒ such as the need to verbally "hand off" aircraft ‒ improves safety by increasing the time controllers have available to focus on separating aircraft.
CAATS is the backbone of Canadian air travel. Connext DDS is the communications backbone of CAATS.
CAATS requires many features unique to Connext DDS. For instance, an air traffic control system processes a huge amount of data. It is not practical to send everything to every station. CAATS makes use of the DDS filtering QoS. For instance, an operator display can request only aircraft tracks within five miles, below 5000 feet, descending, and approaching. RTI Connext DDS sends that specification to the publisher, and filters it on the writer. This only possible because DDS is data centric, so it can understand the contents of the information it conveys. Writer-side filtering, an exclusive Connext DDS feature, prevents sending most of the data that message centric middleware and other DDS implementations would require. This saves bandwidth and enables scalability.
NAV Canada CAATS controls air traffic across the continent. It requires extreme availability, performance and reliability.
The CAATS integrates centers, communicates to external systems and links segments together. Intra-center integration (green boxes) will be operational on Connext DDS in 2015. The external systems will be integrated shortly thereafter.
The US Army must keep track of all its assets. Today, that is done with a system called Blue Force Tracker, or Joint Battle Command - Platforms (JBC-P). Blue Force Tracker must process location information from thousands of devices.
The current system was written with in-house transactional middleware. Transactional systems are very common in the database and financial worlds. AMQP, the Advanced Messaging Queuing Protocol, is an example. They are designed to never lose information. To do that in a distributed system (or a database) requires several messages to start the exchange, to ensure reliable delivery, and then another to confirm the delivery was actually processed.
In this tracking application, however, transactional processing is very expensive. There are many tracks, and updating each one means processing a full transaction. The current system is large — the JBC-P program wrote about 1.5m Source Lines of Code (SLOC) for communications. It can handle about 20,000 tracks on 11 servers with 88 cores. The transactions don't lose messages, but real-time failover is uncertain because it is difficult to quickly switch over to redundant transactional servers.
The Army's new system must handle more than 200,000 tracks. The transactional system couldn't do that, so they switched to a DDS-based design. The new system required only 50k SLOC, because the DDS model handles the communications and, importantly, most of the error handling. By taking advantage of filtering and reliable multicast, it is able to handle 250,000 tracks on 80 percent of a single core. It nonetheless runs on several machines in a private cloud, making it fully redundant. It is also 100 percent standards based.
This example points out the importance of matching the problem and the middleware architecture. The 25x increase in performance by the DDS system is mostly due to its much better fit to the problem.
The US Army's Blue Force Tracker collects information from hundreds of thousands of vehicles over a wide area. The DDS-based system analyzes all the tracks in a private cloud.
Wi-Tronix wirelessly monitors mobile assets such as locomotives, industrial and mining equipment, and marine vessels. It uses wireless technology, including cellular networks, GPRS and infrastructure Wi-Fi local area networks to track and monitor mobile assets throughout the world. Wi-Tronix rail customers include BNSF Railway, Canadian National Railway and the Florida East Coast Railway.
The system keeps track of the positions of the mobile systems wherever they go. It also monitors key parameters like utilization, duty cycle and fuel consumption. By integrating with an on-board DDS bus, it also collects information from engine controllers, vehicle controllers, remote control systems, digital video recorders and black-box data recorders.
RTI middleware must discover data sources and control flow over intermittent wireless networks. DDS QoS control helps immensely. For instance, if a train goes through a tunnel for five minutes, DDS can be configured to expect that and maintain the connection. A TCP socket, by contrast, would disconnect and then face a complex and slow reconnection process.
DDS makes it easier to monitor moving assets over wireless networks. It handles discovery (finding the assets), switching between networks and maintaining connection during lapses in service.