Comparative Analysis of Large Data Transfer in Automotive Applications Using Ethernet Switched Networks

Automotive radars and cameras form the backbone of self-driving cars, Active safety and Advanced Driver Assistant Systems (ADASs). Streaming sensor, camera and audio data between sensors and Electronic control units(ECUs) requires huge data exchange. Ethernet with its large bandwidth capability is typically used as physical medium to communicate large data in automotive in-vehicle networks. Large data generally deals with payload size greater than single Ethernet Maximum Transmission Unit(MTU) size i.e.1522 Bytes that shall be sent via the transport protocol of the underlying bus. The purpose of the paper is to compare four different methods for transferring large data for current automotive Ethernet requirements. The methods Udp/Ip Fragmentation, Application fragmentation, Tcp/Ip Fragmentation and IEEE1722 transport protocol are evaluated. Bench evaluation of these methods are performed using Vector/Elektrobit software. The optimal approach is evaluated based on network performance with End to End latency metric.