History and Development of the Controller Area Network (CAN Bus)

Introduction

The Controller Area Network (CAN bus) is a robust serial communication bus originally developed for automotive applications in the 1980s. It allows microcontrollers and electronic control units (ECUs) in a vehicle or machine to communicate with each other without a central host computer. Since its inception, CAN bus has become one of the most successful network protocols ever, finding use not only in virtually every modern car but also in trucks, ships, industrial machinery, and medical devices​. This essay provides an overview of CAN bus history, key milestones, adoption statistics, major applications, and technological developments from its birth to the present.

Origins and Early Development (1980s)

In the early 1980s, engineers at German automotive supplier Robert Bosch GmbH saw the need for a new in-vehicle network. Existing serial buses could not meet automotive requirements for real-time control and reliability​. Led by engineer Uwe Kiencke, Bosch began developing a custom serial bus protocol in 1983​. The primary goal was to enable new functionalities and coordination between electronic devices in cars; reducing the complexity of wiring was a secondary benefit. By February 1986, Bosch officially introduced the new protocol – named Controller Area Network (CAN) – at the SAE (Society of Automotive Engineers) congress in Detroit​. The CAN bus was designed with an innovative non-destructive arbitration method that let multiple nodes transmit without collision, and it incorporated robust error detection with automatic fault confinement (faulty nodes would disconnect themselves)​.

This early development phase culminated in hardware support. In mid-1987, Intel delivered the first CAN controller chip (the 82526), shortly followed by Philips’ 82C200 CAN controller​. These chips implemented the CAN protocol in silicon, kick-starting the deployment of CAN in real products. The rapid progress – going from concept to working silicon in about four years – showed the immediate promise of CAN bus.

Key early milestones:

• 1983: Development of CAN protocol begins at Bosch under Uwe Kiencke​.

• 1986: CAN formally unveiled by Bosch at the SAE Congress in Detroit, marking the public debut of the technology​.

• 1987: Intel and Philips release the first CAN controller ICs (82526 and 82C200), making CAN readily available for vehicle manufacturers​.

Standardization and Automotive Adoption (1990s)

Following its launch, CAN bus quickly gained support from automakers. Bosch released the CAN 2.0 specification in 1991, defining both standard (11-bit identifier) and extended (29-bit identifier) frame formats​. The protocol then underwent international standardization: in 1993, CAN was adopted as ISO 11898, which standardized the data link layer and a high-speed physical layer up to 1 Mbit/s​. (An additional low-speed, fault-tolerant CAN standard ISO 11519-2 was drafted in the mid-90s, though it saw little use due to technical limitations​.) In 1995, the ISO 11898 standard was updated with an addendum to include the extended 29-bit identifier format, reflecting the CAN 2.0B capability​. By 2003, the ISO 11898 standard had been refined and re-published as a series of standards (11898-1 for the data link layer, 11898-2 for the high-speed physical layer, and 11898-3 for low-speed fault-tolerant CAN)​.

Automakers began installing CAN in production vehicles in the early 1990s. Mercedes-Benz was an early adopter – starting in 1991, Mercedes used CAN to network engine management controllers in its luxury models​. BMW followed suit; for example, the 1995 BMW 7-Series incorporated a CAN bus linking five ECUs in a tree/star topology​. Throughout the 1990s, other manufacturers (especially in Europe) embraced CAN, typically implementing two or more CAN networks per vehicle (for powertrain and body control) with gateway modules linking them​. By the 2000s, CAN bus had become the de facto standard for in-vehicle networking worldwide.

A major catalyst for universal adoption was the implementation of On-Board Diagnostics. In the United States, regulations mandated that by 2008, all new light-duty vehicles sold must support OBD-II over CAN (specifically ISO 15765-4 CAN) for emissions diagnostics. In other words, starting with model year 2008, every car in the US had to have a CAN bus for diagnostic communication​. This mandate essentially cemented CAN’s presence in all cars (most manufacturers had already transitioned to CAN in the early 2000s, but the mandate eliminated any remaining proprietary alternatives for diagnostics). By this time, a modern automobile typically had multiple CAN networks and dozens of CAN nodes – a contemporary car can easily contain 50 to 70+ ECUs (each a CAN node) coordinating via CAN bus​.

In the heavy-duty vehicle realm, CAN was also adopted via the SAE J1939 higher-layer protocol, which was developed in the 1990s for truck and bus networks. J1939 uses the CAN 2.0B extended frame format for communication, and it became the standard for engines, brakes, and other systems in trucks, buses, and off-highway vehicles. This standard was later formalized internationally (e.g. ISO 11992 for truck/trailer communication and ISO 11783 for agricultural machinery are based on J1939 over CAN)​. Thus, by the end of the 1990s, CAN bus was not only standard in passenger cars but was also the dominant network for heavy vehicles.

Automotive adoption milestones:

• 1991: Bosch publishes CAN 2.0 spec; Mercedes-Benz introduces CAN in production vehicles (S-class models)​.

• 1993: CAN bus becomes an international standard (ISO 11898)​.

• 1995: BMW and others integrate CAN for both powertrain and body systems; ISO 11898 updated to include 29-bit ID (extended frame format)​.

• 2003: Revised ISO 11898 series published (defining CAN’s data link and physical layers)​; multiple CAN networks per car are common practice​.

• 2008: All new cars in the U.S. are required to use CAN for OBD-II diagnostics​, marking the full saturation of CAN in the automotive market globally.

Expansion to Industrial and Medical Applications

While born in automotive, CAN bus saw rapid expansion into other industries in the 1990s and 2000s due to its reliability, real-time capability, and low cost. Early on, engineers in various fields recognized CAN’s suitability for embedded control networks:

• Industrial Automation: In the early 1990s, a joint project by Allen-Bradley and others explored using CAN in factory automation, leading to the development of higher-layer protocols on CAN. Allen-Bradley introduced DeviceNet in 1994 (eventually handed over to the Open DeviceNet Vendor Association)​. DeviceNet became a popular fieldbus in North America for connecting industrial sensors, actuators, and controllers using CAN as the underlying transport. Around the same time in Europe, a consortium led by Bosch worked on what became CANopen – a flexible CAN-based application layer for machine control. The CANopen specification was released in 1995 and quickly gained wide adoption in machinery, robotics, and automation across Europe​. Within only a few years, CANopen became a leading embedded network standard in European industries, standardized as EN 50325-4 in 2003. Numerous other niche CAN-based protocols also emerged (e.g. CAN Kingdom for textile machines in the early ’90s​, Honeywell’s SDS for sensors, etc.), further spreading CAN in industrial equipment. By the late 2010s, it was estimated that about 20% of all CAN nodes were used outside the automotive sector – equating to hundreds of millions of CAN devices in factory automation, industrial machines, and other non-automotive applications​. In fact, CAN is often described as “the dominating embedded network in real-time control systems” across many industries​.

• Medical Devices: The medical technology sector also adopted CAN bus for its internal networks. Philips Medical Systems was one of the first manufacturers to use CAN inside medical equipment (networking components within their X-ray imaging machines)​. Philips even developed an early CAN-based application protocol called the Philips Message Specification (PMS) for this purpose​. Following Philips’ lead in the 1990s, other medical device makers incorporated CAN – particularly using the CANopen protocol – to link sensors, motors, and controllers in complex devices. Today, CANopen-based networks are used in a variety of medical equipment: X-ray and CT scanners, MRI-compatible infusion pumps, patient monitors, laboratory analyzers, surgical tables and lights, etc.​. The CAN bus’s noise immunity and error checking make it well-suited for reliable operation in medical electronics. Even some hospital operating rooms have CANopen networks coordinating their automated systems​.

• Other Sectors: Many other fields found uses for CAN. In building automation and elevators, companies like Kone (Finland) deployed CAN networks in their systems as early as the late 1980s​. In maritime and rail, CAN connects subsystems on ships and trains. Aerospace also uses CAN for certain subsystems (e.g. CAN is used in some aircraft for secondary systems and in satellites or launch vehicles for component communication). By the 2010s, CAN bus was truly ubiquitous in embedded systems. As one source summarizes: today CAN bus is standard not only in cars, trucks, and buses, but also in ships, aircraft, off-road machinery, electric vehicles, and industrial machines​. Wherever a durable, real-time field network is needed, CAN has often been the solution.

Growth Trends and CAN Node Statistics

The spread of CAN bus over the decades is reflected in the explosive growth in the number of CAN nodes (devices) produced and in use. In the automotive sector alone, tens of millions of new cars each year, each containing multiple CAN-enabled controllers, drove volume into the hundreds of millions of nodes annually by the turn of the century. Industry groups like CAN in Automation (CiA) have tracked these numbers over time:

• 1990s: In the late 1990s, CAN was still ramping up. Around 1999, annual CAN node production was on the order of tens of millions and growing quickly as more vehicles and machines adopted electronic networks.

• Early 2000s: By the early 2000s, CAN node shipments had reached the hundreds of millions per year. A study by CiA showed that around 2003, roughly 400 million CAN nodes were being sold each year​, up from virtually zero a decade earlier. This rapid rise was driven by mass production of affordable CAN controller chips integrated into microcontrollers for cars and industrial devices.

• Mid 2000s: Growth continued at a high rate. By 2007, an estimated 600 million CAN nodes were in use (cumulative) across industries​. Nearly 80% of these were in automotive applications, with the remainder in areas like industrial automation. Automakers were installing more ECUs per vehicle (for new features and stricter safety/emission requirements), each ECU adding a CAN node.

• 2010s: The proliferation of CAN in all passenger vehicles worldwide, plus continued adoption in trucks, industrial equipment, and other electronics, pushed annual production into the billions. By the late 2010s, industry analysts projected around 1.5 billion new CAN nodes being installed in 2017 alone​. In 2019, just before an industry-wide dip in sales, the annual installed CAN nodes reached approximately 2 billion globally​. Of these, roughly 80% (1.6 billion) were in automobiles and 20% (around 400 million) were in other industries​. These figures illustrate both the dominance of automotive uses and the significant presence of CAN in industrial and commercial products.

• 2020s: As of the early 2020s, the cumulative number of CAN-equipped devices in operation is well into the tens of billions. Virtually every car on the road has multiple CAN nodes, and production of new vehicles (on the order of 70–90 million per year globally) adds further hundreds of millions of CAN nodes annually. Even with some impact from newer networking technologies, CAN continues to ship in enormous volumes each year.

Overall, the trend has been one of steady, significant growth in CAN bus adoption over time. The number of CAN nodes per vehicle also grew: early CAN-equipped cars in the 1990s might have had a handful of nodes, whereas a modern high-end vehicle may contain over 100 CAN-connected modules when counting all the different CAN networks (powertrain CAN, body CAN, infotainment CAN, etc.). This, combined with the expansion of CAN into more product categories, explains the exponential rise of CAN node installations. By 2019, annual CAN node shipments were five times higher than in 2003, highlighting how widely CAN spread in just 15 years​.

Technological Developments Over Time

Throughout its history, the CAN bus standard has evolved to meet new requirements while retaining backward compatibility. Key technological developments include:

• Higher-Layer Protocols: While the CAN standard defines only the lowest network layers (physical and data link), various higher-layer protocols were developed to standardize how CAN is used in different domains. Examples (already mentioned) include DeviceNet, CANopen, J1939, and CAN Kingdom. These enabled plug-and-play interoperability and more complex communication (transport layer, standardized messages) on top of CAN, fueling its adoption in industry and ensuring that CAN could be adapted to diverse applications​.

• Physical Layer Options: CAN was originally implemented as a high-speed two-wire bus (ISO 11898-2) running up to 1 Mbit/s. In the 1990s, automakers also desired low-speed fault-tolerant networks for simpler devices (e.g., door locks, windows). Bosch and others created a low-speed single-wire CAN and other physical variants (ISO 11898-3, ISO 11519-2), although the high-speed two-wire CAN remained most common. Transceiver improvements over time increased robustness against electromagnetic interference and allowed CAN to be used in electrically noisy environments (one reason it’s so popular in industrial and automotive contexts).

• CAN 2.0A vs 2.0B: The initial CAN 2.0 specification included the standard frame format with 11-bit IDs. An extended format with 29-bit identifiers was introduced soon after (formally added in 1995’s update to ISO 11898)​. Modern CAN controllers support both formats, allowing over 500 million unique IDs in a network if needed. This extension made CAN more usable in large systems (like heavy vehicles and complex machines) by providing a bigger address space.

• CAN FD (Flexible Data Rate): By the late 2000s, some applications began to push the limits of Classical CAN (which maxes out at 1 Mbps and 8 bytes of data per frame). To support higher data throughput (for things like advanced driver assistance systems, high-resolution sensors, etc.), Bosch developed an enhanced version of CAN called CAN FD in 2011–2012​. CAN FD increases the maximum payload size from 8 bytes to 64 bytes and allows switching to a higher bit rate during the data phase of the frame (e.g. up to 5 or even 8 Mbps, depending on transceivers), while still arbitrating at the normal 1 Mbps rate​. The CAN FD 1.0 specification was released by Bosch in 2012, and it was standardized in ISO 11898-1:2015 (data link layer) and ISO 11898-2:2016 (CAN FD high-speed physical layer up to 5 Mbit/s)​. CAN FD is largely backward-compatible – Classical CAN nodes see FD frames as extended frames they don’t understand and ignore them – which eased its introduction alongside legacy systems. Automakers began planning migrations to CAN FD for certain networks; the first production cars with CAN FD were expected around 2019/2020​. Indeed, CAN FD is now appearing in vehicles for high-bandwidth functions, though Classical CAN remains prevalent for most networks.

• CAN XL: As a forward-looking step, the CAN in Automation group started work on CAN XL in 2018 – a next-generation CAN protocol intended to support even larger payloads (well beyond 64 bytes) and higher bit rates (approaching 10–20 Mbps). As of 2023–2024, CAN XL was in the final stages of standardization (with ISO 11898-1:2024 and 11898-2:2024 slated to cover it)​. CAN XL adds features to keep CAN relevant for future use cases that may demand more bandwidth (for example, high-resolution sensor data or over-the-air updates in vehicles) while still offering CAN’s core benefits.

Throughout these developments, CAN bus retained its key advantages: low cost, simplicity, reliability, and real-time performance. Competing in-vehicle network technologies were introduced – LIN bus for low-cost body electronics, FlexRay for fault-tolerant high-speed applications, and more recently Automotive Ethernet for high-bandwidth data – but none have displaced CAN. Instead, they often coexist with CAN. For instance, FlexRay saw some use in the 2000s for luxury car drive-by-wire or active suspension systems, but its adoption was limited. Ethernet is now used for infotainment, cameras, or advanced driver assist systems in modern cars, yet CAN still handles the critical real-time control of engines, transmissions, and body functions. Industry experts anticipate that in the coming years, Automotive Ethernet, CAN FD, and CAN XL will all be used in parallel in new vehicle architectures, each for different domains of the car​. In short, CAN bus technology has continually evolved to address new requirements, ensuring it remains a cornerstone of vehicle and machine networking.

Conclusion

Over roughly four decades, the Controller Area Network has grown from an idea in a Bosch lab into a globally ubiquitous network protocol. Key milestones in CAN bus history – its 1986 debut, 1993 standardization, widespread 1990s automotive adoption, extension via CAN FD in the 2010s – mark the timeline of a technology that solved a crucial problem (efficient multiplex communication) and kept advancing. The statistics speak to its success: from zero to billions of nodes per year, CAN’s adoption curve mirrors the rising electronification of modern vehicles and equipment.

Crucially, CAN bus gained traction beyond automobiles, becoming a workhorse in industrial control systems and medical devices due to its robustness and efficiency. Whether inside a car’s engine controller, a factory robot, or a hospital X-ray machine, CAN provides a reliable communication backbone. Its longevity can be attributed to a combination of technical strengths (real-time deterministic behavior, error confinement, etc.) and industry support (standardization and improvements over time). Even as we enter an era of smarter, more connected vehicles and Industry 4.0 automation, CAN is expected to “stay relevant” in the foreseeable future​ – often working in tandem with newer technologies. In summary, the CAN bus has had a remarkable development journey and continues to adapt, maintaining its role as a fundamental network for embedded communication across automotive, industrial, medical, and many other domains.