APPLICATION OF INDUSTRIAL CAN BUS TECHNOLOGY

FOR LEO-SATELLITES
Klaus Janschek, Annerose Braune
Technische Universitt Dresden, Department of Electrical Engineering,
Institute of Automation, D-01062 Dresden, Germany
ABSTRACT
The applicability of the CAN bus technology for cost reduction potentials in the area of onboard
automation of LEO-satellites is investigated. Due to its source from automotive applications the CAN bus
technology is a very promising candidate in terms of functionality, performance, environmental robustness
and cost. It furthermore supports efficiently functional redundancy due to its multi-master capabilities. The
paper analyses the specific features of the CAN bus compared with other industrial field busses and
discusses the functional performances for a combined telecommunication and earth observation reference
mission. A simulation tool is presented which allows to analyse and optimise the bus arbitration design.
1. INTRODUCTION
Low Earth Orbit (LEO) satellites play an increasing commercial role in the area of mobile communication
and earth observation including traffic and environmental monitoring. These LEO-satellites are built under
stringent commercial constraints. Considerable cost reduction potentials in the area of onboard automation
are found through the implementation of functional redundancy to decrease the overall number of
equipment and through the use of well proven industrial (COTS - commercial off the shelf) technologies
for the onboard implementation. Functional redundancy for the onboard automation requires appropriate
bus-technologies for an efficient real-time communication between the different equipment with maximum
freedom for reconfiguration. Due to its source from automotive applications the CAN bus technology is
a very promising candidate in terms of functionality, performance, environmental robustness, real-time
features and cost.
The paper investigates the applicability of the CAN bus technology for a combined telecommunication and
earth observation reference mission which deals with road traffic monitoring (TUD-Satellite [1]). The
underlying onboard data handling and control system has in this case to deal with standard platform and
complex payload data originating from a radio data collecting payload and an earth observation camera.
2. INDUSTRIAL FIELDBUS SYSTEMS
For the above described mission three international field-busses were compared: Profibus DP, Local
Operating Network (LON) and Controller Area Network (CAN). Table 1 shows some results. In principle
all three field-busses meet the requirements for LEO satellite applications, but differ in details. The master
slave bus access method of Profibus DP demands a redundant bus master. LON is used mainly for building
services automation and supports therefore a lot of - in our case not necessary - network and routing
functionality for distributed networks up to some 100 m bus length. The special advantage of CAN is the
design for robust environmental and reliable application in the automotive industry.
2
nd
IAA Symposium on Small
Satellites for Earth Observation,
Berlin, Germany, April 12 16,
1999
Table 1: Comparison of several industrial field-busses
Criterion Requirement Profibus DP LON CAN
No. of nodes
Bit Rate
max. 20
> 1Mbit/s
max. 32
12 Mbit/s
max. 64
1,25 Mbit/s
max. 64
1Mbit/s
Hardware Environ.robustness partly partly typical for auto-
motive industry
Length @ max bit rate max. 40 m 40 m 500 m 40 m
Telegram flexible word size
high data security
partly (0 .. 32 byte)
high (HD=4, parity)
partly (0 .. 228 byte)
high (16bitCRC,HD=4)
partly (0 .. 8 byte)
high (16 bitCRC,HD=6)
Protocol access procedure with
guaranteed response
Master-Slave
yes
multi-master
no
multi-master
for high prior mess.
Economical Condition
Cost
international
standardisation
less than space qualif.
DIN EN 50170
yes
no/ at the market
yes
ISO/DIS 11519
ISO/DIS 11898
yes
3. SATELLITE APPLICABILITY ANALYSIS FOR CAN-TECHNOLOGY
3.1 Standardisation and availability
Due to the use in automotive industry a lot of cost effective CAN controllers exist with different
characteristics. CAN controllers, operating directly in the engine compartment of a car, correspond already
nearly to the requirements of LEO satellites with regard to temperature range and mechanical robustness.
So they are already well suited for space applications. With CMOS-technology also the radiation
robustness of the low orbit can be met to a large extent. New
developments offer e.g. integrated sensors with included CAN-
controllers applicable directly as navigation sensors [2],[3].
A considerable advantage of industrial field-bus technology is
its widely spread international standardisation. According to the
OSI reference model (see Fig.1) the CAN architecture
incorporates two layers: data link layer and physical layer. CAN layers 1 and 2
are international standards. So we use the defined interface to
layer 2 for our application.
On this basis almost every commercial software development
environment (compilers) have available well tested CAN-
software components, which offers enormous potential for cost
reduction of onboard software development.
3.2 Data handling
One CAN-message transmits up to 64 Bit non structured user data (see Fig.2). This allows the
transmission of device or process system data with one or only a few messages. Data blocks have to be
structured into 64 bit units. There are no services for structured data or block transfer. CAN supports a
max. bit rate of 1 Mbit/s at a max. 40 m bus length. Assuming 70 % bus capacity utilisation, will allow
a maximum user data rate
of 400 Kbit/s, which meets
most small satellite platform
data requirements.
7 Application
6Presentation
Application
Layer
DataLink
5Session
4Transport
3 Network
2DataLink
1 Physical
Line/Driver
Figure 1: CAN versus ISO/OSI
Reference Model
98 ... 106
b
user data
max. 64
checksum
15 bit
....
identifier
11 bit
....
Figure 2: CAN data frame
start
bits
3.3 Multi master bus access
A rigid Master-Slave protocol access procedure is not suitable for an autonomous satellite application
because of the necessary redundancy of the bus master. CAN uses a multi-master, priority-based bus access
with a modified CSMA/CD access method. All nodes are operating as masters. When the bus is idle, any
node may start to transmit a frame. If two or more nodes start to transmit frames at the same time, the bus
access conflict is resolved by contention-based arbitration using an 11-bit identifier. The mechanism of
arbitration guarantees that no information and no time is lost. The transmitter with the frame of highest
priority receives bus
access. So the protocol already realises inherently flexible communication structures and broadcasting
capabilities. Frames, that have lost arbitration because of lower priority will be retransmitted automatically
when the bus is idle again. The resulting time delay depends on the other higher prior frames. So a
predictable time delay is only possible for highest prior frames. Nodes may be added to the CAN network
without requiring any changes in the software or hardware.
3.4 Error detection, isolation and recovery
The advantages of the CAN results of his automotive application area and the demanding requirements for
error detection, isolation and recovery. Corrupted frames will be detected by any transmitting nodes and
any nominally operating receiving node. Such frames will be aborted and retransmitted according to the
implemented recovery procedure. CAN nodes can distinguish short disturbances from permanent failure.
A node can have several error states with different access rights. Unfortunately if completely defect, no
more transmitting nodes are not automatically detected at layer 2. Therefore user defined services are
necessary.
4. REAL TIME BEHAVIOUR ANALYSIS OF A REFERENCE ONBOARD STRUCTURE
Paragraph 3 shows some critical aspects especially concerning the real time behaviour of the CAN.
Therefore some algorithms were tested for analysing the real time characteristics. Basis for the tests is a
reference onboard configuration (nodes, incl. the awaited frames with their cyclic time, user data length,
permissible deadline) as shown in Figure 3. For cyclic frames the expected bus load can be determined
(frame length/cyclic time for all frames). Under the assumption of a bus load lower than 100%, it is
possible to define with special algorithms the identifier for every frame and calculate the real worst case
deadline [4],[5]. If all real deadlines are lower than the permissible ones, the onboard configuration meets
the real time requirements else better suited identifiers have to be found.
This real time analysis and a follow-on bus simulation to verify and monitor the results were implemented
with the simulation tool STATEMATE .
The following assumptions have been used for a first test of real time behaviour the following onboard
configuration (see Fig 3. ):
redundant CAN bus
platform attitude and orbit control equipment
payload: Earth observation camera, RF road traffic data collection equipment
with several nominal and failure modes.
The following results have been found for this configuration:
- bus capacity for platform data : about 15%,
- bus capacity for payload data: about 60%.
- Some single node error can be tolerated ( e.g. GPS pre-processing).
- In case of failure of one CAN-bus the payload data rates have to be reduced.
In particular no critical deadline overruns could be detected. So we can assume, that in principle a
redundant CAN is suited for this example.
5. SPACE ROBUST CAN CONTROLLER
The critical issue for the use of onboard CAN systems is the radiation robustness of the controllers.
Although in certain LEO environments standard CMOS-technology has shown to operate appropriately,
it would be interesting, to have available also components showing increased radiation robustness. In this
area we are currently performing feasibility investigations in co-operation with the Frauenhofer
Gesellschaft (FhG) Institute for Microelectronics, Dresden. The investigations intend to reuse already
existing CAN chip design solutions at FhG for automotive applications and to adapt them slightly in terms
of functionality and production process characteristics [6].
6. SUMMARY
The investigations show, that CAN bus technology is a very promising candidate for the LEO satellite
application due to its basic concepts: multi-master priority bus access, non-destructive contention-based
arbitration, multicast frame transfer by acceptance filtering, configuration flexibility, system wide data
consistency, error detection and signalling; distinction between temporary errors and permanent failures
of nodes and autonomous switching off of defective node.
The standardised industrial controller or modules promote a cost effective design and implementation of
failure redundant onboard data handling structures.
7. REFERENCES
[1] Janschek, K., Scharfe, M.: The TUD-Satellite Demonstration Mission Satellite-Based Monitoring of Mobile Objects,
Proceedings of the CNES International Conference on Small Satellites, Systems & Services, Antibes, France, September
1998
[2] Ulbricht, S; Budde, W.; Gottfried-Gottfried, R.; Sauer, B.; Wende, U.;A monolithically integrated two-axis magnetic field
sensor system,ESSCIRC,1996, Proceedings pp. 132-135.
[3] Bender, S.; Fischer, W.J.; Heinig, A.; Khler, C.; Rietz,S.; Schneider,B.: Multi-sensor controller with CAN bus interface,
Micro system technologies, 1996, Potsdam, Proceedings pp. 151-155.
[4] Staub, M.: Mit dem Worst Case kalkuliert - Zugriffsgarantie bei CAN-Netzen, elektronik 1995, Heft 12.
[5] Stmpfle, M.: Ober sticht unter - Priorittenvergabe in Echtzeitsystemen, elektronik 1995, Heft 22.
[6] Janschek, K., Boge, T., Braune, A.: Einsatz industrieller Systemkomponenten in Kleinsatelliten am Beispiel des CAN-
Busses. Vortrag, 8. Avionik-Workshop Avioniksysteme fr die Raumfahrt (Spaceionik), DGLR FA Avionik (T6.2),
28.Oktober 1998, Garching.
Figure 3: Onboard reference configuration
reaction
wheel
magnet
torquer
magnet
torquer
magneto
meter
magneto
meter
magneto
meter
onboard
computer
Camera
TM/TC
GPS
magnet
torquer
preprocessing