Distributed Electrical/Electronic Architecture (Distributed EEA)

Distributed Electrical/Electronic Architecture (Distributed EEA) is an E/E system design approach widely adopted during the early stages of automotive intelligence. Its defining characteristic is "one function, one control unit"—where every electronic function of the vehicle, from power window operation to engine management, is governed by an independent Electronic Control Unit (ECU).

Technical Working Principles and Architectural Characteristics

Under a distributed EEA, dozens or even hundreds of ECUs are located throughout the vehicle, interconnected via conventional in-vehicle bus networks such as CAN, LIN, and FlexRay to handle essential signal exchange and control.

Hardware Level: Each function corresponds to a dedicated ECU containing its own microcontroller, power management circuitry, casing, and interface circuits. A modern mid-to-high-end vehicle can feature between 70 and 100 ECUs, resulting in a total wiring harness length of 2 to 5 km and a weight exceeding 50 kg.

Software Level: Software functions are hardcoded into the hardware of their respective ECUs, creating a tightly coupled hardware-software setup. These ECUs are typically developed independently by different Tier-1 suppliers, each running disparate software frameworks and underlying code without a unified software architecture. Communication relies on a Signal-Oriented approach, where ECUs broadcast data periodically or upon changes, regardless of whether recipient nodes actually require it.

Historical Role and Limitations

Distributed EEA supported the automotive industry’s smooth transition from mechanical systems to electronics over recent decades. Its key strength lies in the relative independence of each subsystem, ensuring that a fault in a single controller will not compromise the operation of the entire vehicle. However, as modern vehicles become increasingly intelligent, its inherent drawbacks have become ever more pronounced.

Inefficient Computing Power Utilisation: To ensure operating stability, each ECU is designed with surplus processing headroom; however, looking at the vehicle as a whole, this fragmented computing power cannot be effectively coordinated or shared. For context, typical Level 2+ (L2+) autonomous driving systems demand over 200 TOPS of processing power, whereas distributed ECUs typically run on basic MCUs clocked at just tens of MHz, falling far short of these demands.

Communication Bandwidth Bottlenecks: A standard CAN bus provides a bandwidth of only around 1 Mbps, while CAN-FD pushes this to 8 Mbps—still insufficient for high-bandwidth data streams like those from ADAS cameras. An 8-megapixel camera, for instance, generates a raw data stream of roughly 2 Gbps, vastly outstripping the transmission capacity of CAN bus systems.

Complex OTA Updates: With tightly coupled hardware and software sourced from various suppliers, rolling out functional upgrades requires modifying individual ECUs one by one, making full-vehicle Over-The-Air (OTA) updates virtually impossible.

Ballooning Wiring Harness and Costs: Adding any new feature necessitates dedicated ECUs and additional wiring harnesses, which not only drives up bill-of-materials (BOM) costs but also significantly increases manufacturing and assembly complexity.

Current Industry Landscape and Evolution

According to the E/E architecture roadmap mapped out by industry leaders like Bosch, distributed architecture represents the first-generation E/E architecture, broadly categorised into two phases: "modular" and "integrated". The industry is currently transitioning from distributed setups towards domain-centralised and central computing architectures, with the first major wave of this architectural migration expected to conclude around 2025. While distributed EEA remains prevalent across existing legacy models, domain-centralised and central computing architectures are rapidly becoming the standard for the new generation of intelligent vehicles.