1 Introduction
Modern vehicles have become increasingly computerized to satisfy the more strict safety requirements
and to provide better driving experiences. Therefore, the number of electronic control units
(ECUs) in modern vehicles has continuously increased in last few decades (e.g., some luxury cars
can have up to 100 ECUs). In addition, advanced functionalities (e.g., collision avoidance and
adaptive cruise control) put higher computational demand on ECUs, which further increases the
design complexity of automotive control systems. Multicore processors, where multiple processing
units are integrated on a single chip, have emerged to be the main computing engine not only for
high-end servers but also for embedded control systems. For instance, some multicore processors
have been recently developed for automotive ECUs [3, 11].
With multicore processors, more centralized architecture designs can be adopted for automotive
control systems. That is, instead of having many ECUs following the traditional approach of “one
function per ECU”, we can have a few powerful multicore ECUs and each of them integrate the
functionalities of several single-core ECUs from the same or similar domains (e.g., powertrain and
body). The recent initiative on AUTOmotive Open System ARchitecture (AUTOSAR) has established
several standards for automotive software and hardware designs, which include guidelines
for designing centralized architecture with multicore ECUs for automotive control systems [2, 5].
With AUTOSAR, it is expected that computational control tasks of different functions can share
one ECU or run on any ECU connected with in-vehicle network (e.g., CAN and FlexRay).
Although multicore processors provide great opportunities to mitigate the design complexity
of automotive control systems with reduced number of ECUs, the integration and scheduling of
different control tasks on multicore ECUs also bring various challenges. In what follows, we first
discuss the challenges for designing the hardware architecture with multicore ECUs. Then, we will
address the benefits and scheduling issues within multicore ECUs and across in-vehicle networks.
2 Multicore Domain Control Units (MDCUs)
Following the “one function per ECU” design paradigm, the number of ECUs in modern vehicles
is typically around 50 to 70, ranging from basic safety-related functions (e.g., ABS and engine and
transmission control) to auxiliary and advanced functions (e.g., navigation and collision avoidance
systems) [1]. With increased computation power in modern processors, the control tasks from several ECUs can be integrated and processed on a single powerful ECU. For instance, engine and transmission control units have been merged to be the powertrain control module [1]. With the rich
computation power of multicore processors, it can be expected that more control tasks from the same or similar domains will be integrated and run on a multicore domain control unit (MDCU).
Therefore, instead of having nearly 100 ECUs, depending on the integration levels, future vehicles
may have around 10 to 20 MDCUs assuming that each MDCU has 4 to 8 processing cores, which
are connected through CAN or FlexRay network, as illustrated

DOWNLOAD FULL REPORT
http://googleurl?sa=t&source=web&cd=1&ve...0Dakai.pdf&ei=aasBTrXCL-3XiAK7lrSoCA&usg=AFQjCNHcvDhWUn4JumiEbKsEND49RSSgOQ&sig2=DJMP68dv9Q-ScAWfg8z_Ng