Intelligent Automation & Soft Computing

ISSN: 1079-8587 (Print) 2326-005X (Online) Journal homepage: http://www.tandfonline.com/loi/tasj20

Performance Evaluation of Node-mapping-based

Flexray-CAN Gateway for in-vehicle Networking
System

Man-Ho Kim, Suk Lee & Kyung-Chang Lee

To cite this article: Man-Ho Kim, Suk Lee & Kyung-Chang Lee (2015) Performance Evaluation
of Node-mapping-based Flexray-CAN Gateway for in-vehicle Networking System, Intelligent
Automation & Soft Computing, 21:2, 251-263, DOI: 10.1080/10798587.2014.981999

To link to this article: http://dx.doi.org/10.1080/10798587.2014.981999

Published online: 20 Nov 2014.

Submit your article to this journal

Article views: 36

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tasj20

Download by: [Gazi University] Date: 16 April 2016, At: 06:18

 Intelligent Automation and Soft Computing, 2015
Vol. 21, No. 2, 251–263, http://dx.doi.org/10.1080/10798587.2014.981999

PERFORMANCE EVALUATION OF NODE-MAPPING-BASED FLEXRAY-

CAN GATEWAY FOR IN-VEHICLE NETWORKING SYSTEM

MAN-HO KIM1, SUK LEE2, AND KYUNG-CHANG LEE3*

1
Leader in Industry-University Cooperation, Pusan National University, 63 Geumjeong-gu, Busan
609-735, Korea
2
School of Mechanical Engineering, Pusan National University, 63 Geumjeong-gu, Busan 609-735, Korea
3
Department of Control and Instrumentation Engineering, Pukyong National University, 365 Sinseon-ro
Nam-Gu, Busan 608-739, Korea
Downloaded by [Gazi University] at 06:18 16 April 2016

ABSTRACT—As vehicles become more intelligent, in-vehicle networking (IVN) systems, such as the
controller area network (CAN) and the FlexRay networks, are essential for the convenience and safety of
drivers. To expand the applicability of IVN systems, attention is currently being focused on the
communication between heterogeneous networks such as body or chassis networking systems. The
message-mapping-based gateway was developed to improve communication between FlexRay and
CAN networks concerning vehicle information. However, there are obstacles to the wide acceptance of
the FlexRay-CAN gateway for a vehicle. First, when the message ID for the network is changed, the
gateway must be reloaded with the revised message-mapping table. Second, if the number of messages
exchanged is increased in the network, the complexity of the gateway software rapidly increases.
In order to overcome these obstacles, this paper presents a FlexRay-CAN gateway using a node-
mapping method. A gateway operation algorithm is described, and an experimental evaluation for ID
change and software complexity is presented.

Key Words: In-vehicle networking (IVN) systems; FlexRay; CAN; Node-mapping-based FlexRay-CAN
gateway; Software complexity; Maximum transmission delay

1. INTRODUCTION

Recently, high transmission rates and predictable real-time transmission delays for in-vehicle networking
(IVN) systems have become necessary to satisfy increasing driver concerns over the convenience and
safety of their vehicles (Kelkar & Kamal, 2014; Taylor et al., 2014; Schmidt, 2014; Han & Han, 2014).
In order to overcome these problems of IVN systems, FlexRay protocols have been developed for chassis
networking systems, with real-time features for braking and steering systems. Automotive vendors have
been attempting to expand the applicability of FlexRay (Park & Sunwoo, 2011; Ouedraogo & Kumar,
2014). In particular, x-by-wire systems are being developed for the FlexRay network. These systems are
replacing rigid mechanical components with dynamically configurable electronic elements and digital
communication networks (Park & Sunwoo, 2011). In contrast, the controller area network (CAN) has a
lower speed and a non-deterministic property of transmission delay. Nevertheless, CAN is widely used in
non-real-time body networking applications such as headlights and power windows (Tindell, Burns, &
Wellings, 1995; Seo et al., 2013).
Today’s highly advanced intelligent vehicles require the exchange of various data between the chassis
and body networks during normal operation (Baronti et al., 2011; Li, Zhang, Xia, & Huang, 2009). Safety

*Corresponding author. Email: [email protected]

q 2014 TSIw Press

 252 Intelligent Automation and Soft Computing

data is frequently exchanged between these networks and the amount of such data exchange is likely to
increase (Lee, Oh, & Kang, 2011). The FlexRay-CAN gateway that transfers messages between the
FlexRay-based chassis network and CAN-based body network enables the two protocols to share data
(Zhao, Qin, & Liu, 2010).
Several studies have discussed data sharing between the FlexRay and CAN protocols for automotive
applications. Zhao et al. (2010) and Schmidt, Alkan, Schmidt, Yuruklu, and Karakaya (2010) studied the
implementation of a FlexRay-CAN gateway using the message-mapping approach. Seo, Ki, Hwang, Kwon,
and Jeon (2009), Chen and Zhang (2011), and Dong, Piao, Jang, Chung, and Lee (2012) also suggested the
implementation of a FlexRay-CAN gateway based on a message-mapping method. Schmidt et al. (2010)
and Qin, Feng, and Xiuwei (2010) utilized a commercial microcontroller unit (MCU) to implement a
message-mapping-based FlexRay-CAN gateway, which maintained an internal table that stored CAN
message IDs and the corresponding FlexRay slot numbers, and utilized the table to convert the received
CAN or FlexRay message to the mapping FlexRay slot or CAN ID. Most research on FlexRay-CAN
Downloaded by [Gazi University] at 06:18 16 April 2016

gateways has focused on the message-mapping approach that provides one-to-one mapping of a FlexRay
and CAN message transmitted via the gateway.
The message-mapping-based FlexRay-CAN gateway has two drawbacks with regards to FlexRay and
CAN network scheduling. These drawbacks are more problematic during the node design phase that
determines the detailed function of each node on the network, rather than during the clustering design phase
that determines the structure of the network systems. First, the number of CAN and FlexRay messages to be
shared increases for message-mapping-based FlexRay-CAN gateways, as the number of FlexRay and CAN
messages to map increases, so that the network-scheduling software is much more complex during the node
design phase. That is, as the number of messages exchanged via the gateway increases, the complexity of
the network-scheduling software that provides the gateway functionality rapidly increases. This leads to an
increased amount of work to be done during the node design phase. In addition, since FlexRay and CAN
network scheduling are configured in the off-line state, it is very difficult to exchange nodes or messages.
In general, the message-mapping-based FlexRay-CAN gateway must modify its message-mapping
information when messages exchanged between the two networks are added or the CAN message IDs are
changed. That is, to add or change messages exchanged via a message-mapping-based FlexRay-CAN
gateway, the operation of the networks must be stopped and the message-mapping information of the
gateway must be reloaded.
This paper presents a node-mapping-based FlexRay-CAN gateway that overcomes the drawbacks of
the conventional message mapping-based FlexRay-CAN gateway without network performance
degradation. The structure and operation algorithm of the proposed gateway are presented, and the
node-mapping method, which assembles and disassembles the messages received by the gateway, is
described. Finally, to verify the feasibility of the proposed gateway, its performance was evaluated using an
experimental model that simulates IVN systems.
The remainder of this paper is organized as follows: Section 2 describes the structure and operation
algorithm of the proposed node-mapping-based FlexRay-CAN gateway. In Section 3, the implementation
of the proposed gateway is described and its performance is evaluated. Finally, conclusions and future
research directions are given in Section 4.

2. STRUCTURE OF NODE-MAPPING-BASED GATEWAY

Figure 1 shows the structure of the node-mapping-based FlexRay-CAN gateway proposed in this paper.
The gateway consists of five modules: A FlexRay transceiver, a CAN transceiver, a FlexRay controller, a
CAN controller, and a gateway controller. First, the FlexRay and CAN transceivers generate message
signals that are transmitted over a network. Second, the controllers are responsible for ensuring that the
 Performance Evaluation of Node-Mapping-Based Flexray-CAN Gateway for In-vehicle Networking System 253

FlexRay-CAN gateway

application

gateway controller

assemble

FlexRay controller CAN controller

CAN node 1 CAN node n

transceiver transceiver
Downloaded by [Gazi University] at 06:18 16 April 2016

CAN network

FlexRay network

FlexRay node 1 FlexRay node 2 FlexRay node n

Figure 1. Structure of Node-mapping-based FlexRay-CAN Gateway.

received message is encoded for transmission and decoding, and that the received code signal is
transformed into the appropriate digital signal. The gateway controller assembles CAN messages into a
FlexRay message or converts the received FlexRay message into a CAN message. That is, when CAN
messages are received from the CAN network, the gateway controller assembles the received CAN
messages according to the assembly procedure in the operation algorithm so as to create a FlexRay message
that is transmitted over the FlexRay network. The gateway controller functions to temporarily store the
received messages in the message buffer, in addition to the message masking and filtering function that
selectively chooses a message from the received messages.
In order to transmit messages over the FlexRay and CAN networks, the proposed node-mapping-based
FlexRay-CAN gateway employs a conversion rule using the node-mapping mechanism shown in Figure 2.
The node-mapping mechanism defines how to assemble and disassemble FlexRay and CAN messages.
More specifically, the FlexRay-CAN gateway assembles CAN messages into a FlexRay slot until the
collected CAN messages exceed the size of the allocated FlexRay slot or when the FlexRay cycle time has
expired. The payload segment of a FlexRay message consists of the following indicators for each of n
messages: number of messages (NoM), start message delimited (SMD), length of message (LoM), and
identifier (ID). NoM is a three-bit indicator representing the number of CAN messages assembled in the
FlexRay message, SMD is a three-bit indicator used to denote the start of the CAN message in the payload,
LoM is a three-bit indicator representing the length of the CAN message and enables a fixed-size gateway
message to contain as many CAN messages as possible, and ID is a 29-bit identifier of the CAN message as
supported in CAN 2.0B. The message segment has a variable length, and provides space for holding the
CAN message data. The length of the message can be extended up to 8 bytes.
In the transmission from the FlexRay to the CAN networks, the gateway removes the header and trailer
of the received FlexRay message, converts it to a CAN message, and stores the converted CAN message in
 254 Intelligent Automation and Soft Computing

node 1 node 2 … node n

CAN network

NoM SMD LoM ID 1 message 1 SMD LoM ID 2 message 2 … SMD LoM ID n message n

header payload trailer

gateway
Downloaded by [Gazi University] at 06:18 16 April 2016

slot G slot 1 … slot n

static segment dynamic segment NIT

FlexRay network

node 1 node 2 … node n

Figure 2. Communication Structure of Node-mapping-based FlexRay-CAN Gateway.

the CAN message queue. The FlexRay-CAN gateway uses the carrier sense multiple access with non-
destructive bitwise arbitration (CSMA/NBA) mechanism to transmit the CAN messages to the CAN
network according to their order in the queue. Indicator information (NoM, SMD, LoM, CAN ID) as well
as message data are assembled into the FlexRay message at the application layer when a FlexRay node
transmits the message to the CAN network. When the FlexRay-CAN gateway receives a FlexRay message
that had been scheduled to be received, it removes the NoM, SMD, and LoM indicators from the frame and
transmits the CAN ID and data to the CAN network.
While, the conventional message-mapping-based FlexRay-CAN gateway allocates a FlexRay slot for
each CAN message (one-to-one mapping). When the gateway receives a CAN message to transmit, the
gateway puts the received CAN message in the FlexRay slot allocated for this message and sends the
message at the allocated time. For example, a CAN message having ID 0x80 is transmitted from the CAN
network, and the gateway transmits FlexRay slot 80 to receive the message in FlexRay slot 80 (the pre-
allocated slot for message 0x80) at the allocated time. In the message-mapping approach, a FlexRay
message is directly mapped to a CAN message, and it is difficult to transmit if the CAN ID is changed.
In addition, if the gateway has not received any CAN messages, empty FlexRay messages are transmitted.
The complexity of the conventional message-mapping-based FlexRay-CAN gateway software increases
rapidly as the number of messages transmitted via the gateway increases. This can slow down the overall
traffic of the CAN and FlexRay networks.
Figure 3 presents the operation algorithm of the proposed node-mapping-based FlexRay-CAN
gateway. After initializing the gateway, it checks the message buffers to receive messages from the
 Performance Evaluation of Node-Mapping-Based Flexray-CAN Gateway for In-vehicle Networking System 255

initialize FlexRay-CAN gateway

no
capture message ?
yes CAN msg. yes FlexRay msg.

store CAN message to queue store FlexRay message to queue

yes
no no
exist CAN message ?
Queue is full ?
yes
no
cycle interval expired ?
Downloaded by [Gazi University] at 06:18 16 April 2016

remove FlexRay format

yes

assemble as one FlexRay message store CAN message to queue

Transmit FlexRay message Transmit CAN message

queue is empty ?
no

yes

Figure 3. Structure of Node-mapping-based FlexRay-CAN Gateway Operation Algorithm.

FlexRay and CAN networks. If the gateway receives a message from the CAN network, the gateway stores
the received message in the message queue. The gateway checks whether the FlexRay cycle time that is set
according to network characteristics has expired. If the FlexRay cycle time has not expired, the gateway
checks the fullness of the message queue. If the message queue is not full, the gateway checks again to see
if there is a message to receive. When the message queue is full, the gateway checks whether the FlexRay
cycle time has expired. If the FlexRay cycle time has expired, the gateway assembles the CAN messages
stored in the message queue into a FlexRay message, and transmits the FlexRay message. If the gateway
receives a message from the FlexRay network, it stores the received FlexRay message in the queue. The
gateway checks if there are any CAN messages to transfer in the stored FlexRay message. If there are any
CAN messages to transfer, the header and trailer of the received FlexRay message are removed and the
FlexRay message is converted to a CAN message. The converted CAN messages are stored in the message
queue and then transmitted to the CAN network sequentially. If there are no further CAN messages to
transmit in the message queue, the gateway checks again to see if there is a message to receive from the
FlexRay and CAN networks.
Figure 4 shows a timing diagram that illustrates message transmissions from the CAN to the FlexRay
networks using the proposed node-mapping-based FlexRay-CAN gateway. In the figure, the generation
cycle of CAN message x1 (mx1) and message y1 (my1) at CAN source node x and y is same as the FlexRay
cycle time. The generation cycle of message x2 (mx2) is twice the cycle time. When the greatest common
divisor of the generation cycle of the CAN messages is set to the FlexRay communication cycle, messages
 256 Intelligent Automation and Soft Computing

source x, y gateway g destination z

message x1
message x2

message y1

time delay
…
message

…
message x1 x1, x2, y1
Downloaded by [Gazi University] at 06:18 16 April 2016

message y1

generation
period

message x1
…

communication cycle
message x2

message y1
message
x1, y1
…

Figure 4. Example Timing Diagram for Node-mapping-based Gateway.

mx1 and my1 are transmitted in slot g at every cycle time, and mx2 is transmitted in slot g at every other
cycle time. When the CAN source node x and y generate and transmit CAN messages x1, x2, and y1, the
CAN ID mask of the FlexRay-CAN gateway verifies that x1, x2, and y1 are messages to receive;
the gateway then receives these messages through the CAN Rx pin. After CAN messages are received, the
gateway adds indicator information (NoM, SMD, and LoM) to the received messages to form FlexRay
message g in which several CAN messages are assembled. When the time to transmit slot g has arrived
during the FlexRay cycle time, the gateway transmits the message to the FlexRay network. Because all
nodes in the FlexRay network obtain network-scheduling information, the CAN messages assembled in
message g are safely delivered to the receiving FlexRay node. In addition, transmission jitter rarely occurs
for the proposed gateway because CAN messages transmitted in FlexRay message g are periodically sent to
the FlexRay network. The conventional message-mapping-based FlexRay-CAN gateway transmits
CAN messages at the same point in the cycle time as the proposed node-mapping-based FlexRay-CAN
gateway, but the difference is that the pre-allocated slot for each CAN message is used for transmission.
The proposed gateway overcomes the drawbacks of the conventional gateway without increasing
transmission jitter.
The proposed FlexRay-CAN gateway transmits messages using the node-mapping mechanism so
that the gateway software does not have to be modified when the message is changed. The transmission of an
empty FlexRay slot is unlikely to occur unless no CAN message is received within the FlexRay cycle time.
The node-mapping mechanism used for message transmission is simple, so the software complexity of
the proposed gateway is relatively low compared to the conventional message-mapping-based FlexRay-
CAN gateway.
 Performance Evaluation of Node-Mapping-Based Flexray-CAN Gateway for In-vehicle Networking System 257

3. PERFORMANCE EVALUATION OF NODE-MAPPING-BASED GATEWAY

Figure 5 shows the experimental testbed used to evaluate the performance of the proposed node-mapping-
based FlexRay-CAN gateway. Figure 5(a) shows a schematic of the experimental testbed upon which the
vehicle body and chassis networks were designed, which was based on the Society of Automotive
Engineers International (SAE) ’Benchmark’ network and message database (Tindell & Burns, 1994). The
experimental testbed consisted of six CAN nodes, six FlexRay nodes, one CANoe.FlexRay 7.2, and one
FlexRay-CAN gateway. The network messages were designed with 29 CAN messages and 30 FlexRay
static slots with 5 ms cycle time by using the 53 message types of the SAE Benchmark. The size of the
FlexRay static slot was set to 64 words. For example, if nine CAN messages, each of which contains 8 bytes
of data, are transmitted, 58 words of the 64-word FlexRay slot in the gateway are filled with the CAN
messages (103 bits or about 13 bytes per CAN message). SAE’s Benchmark network has relatively low
traffic, so more nodes were added to the FlexRay and CAN networks. In addition, the experimental testbed
Downloaded by [Gazi University] at 06:18 16 April 2016

was designed to be able to adjust the message transmission cycle from 5 to 1000 ms.
Figure 5(b) shows the implemented experimental testbed. Freescale’s MC9S12XF512 embedded with
the FlexRay and CAN controllers was used to implement the network nodes (Freescale Semiconductor, Inc,
2010). NXP’s PCA82C250 was used for the CAN transceiver and Austria Microsystems’ AS8221 was used
for the FlexRay transceiver. Vector’s VN 7600 connected to a PC was used to monitor the networks, and
CANoe.FlexRay 7.2 software was installed in the PC to monitor messages transmitted over the networks
(Vector, Inc, 2010). The transmission speed of the CAN network was set to 500 kbps, and that of the
FlexRay network was set to 10 Mbps. The conventional message-mapping-based FlexRay-CAN gateway
was also implemented in the testbed so that it could be compared with the proposed node-mapping-based
FlexRay-CAN gateway.
Three tests were performed to evaluate the performance of the proposed gateway. First, the CAN
message ID was changed; whether or not that message was successfully delivered to the target FlexRay
node was examined. This test was performed to investigate whether the proposed gateway could transmit a
CAN message with a changed ID without having to modify the gateway software. An oscilloscope was
used to monitor the transmitted messages.
The second test examined the software complexity of the proposed gateway. Software complexity was
measured using two different software metrics: Line of code (LoC) and Halstead’s software science (HSS).
LoC measures the complexity of software based on the code length and is the simplest and most widely
used estimation method in the field (Sheng & Shijie, 2010). HSS measures software complexity based on

(a)
IM/C_C battery_C brake_C

CAN network

(b)
vehicle_C driver _C trans._C
CAN Nodes

CANoe.
gateway
FlexRay
CANoe.FlexRay
gateway
IM/C_F battery_F brake_F

FlexRay network

oscillsocope

vehicle_F driver_F trans._F FlexRay Nodes

Figure 5. Testbed for Testing Node-mapping-based Gateway. (a) Schematic Diagram of the Testbed,
(b) Details of the Testbed Implementation.
 258 Intelligent Automation and Soft Computing

Table I. Performance Using the LoC Index.

Conventional message-mapping-based gateway Proposed node-mapping-based gateway

LoC 681 lines 268 lines

the number and frequency of operators and operands in the program. The advantage of HSS is that it can be
applied to any kind of program (Tu, Sun, & Zhang, 2009). HSS can be expressed as follows:
N ¼ N1 þ N2
V ¼ N £ log 2 ðn1 þ n2Þ
ð1Þ
2 £ n2
D ¼ n1 N2

E ¼D£V
Downloaded by [Gazi University] at 06:18 16 April 2016

where N, V, D, and E denote software length, software volume, software difficulty, and software effort,
respectively. And, n1 and n2 denote the number of operators and operands that appear in a program, and N1
and N2 denote their frequencies of appearance.
Table I shows the software complexity of the conventional and proposed gateways as represented by
the LoC. The main and FlexRay configuration codes were counted for the LoC measure. The code of the
proposed gateway was over 60% shorter than that for the conventional gateway. Table II presents the
software complexity of the conventional and proposed gateways in terms of the HSS. The proposed
gateway had a lower HSS value for every measure (i.e., software length, software volume, software
difficulty, and software effort). That is, compared to the conventional gateway, the proposed gateway had
relatively low software complexity. In particular, the software effort measure of the proposed gateway was
only 7% of that of the conventional gateway. Both the LoC and HSS measures indicated that the proposed
gateway had lower software complexity than the conventional gateway.
Finally, the third test examined the transmission delay of the proposed gateway. Here, the transmission
delay was defined as the difference between the time when a message was transmitted through the FlexRay-
CAN gateway and the time when the message was monitored at the CANoe.FlexRay 7.2. Figure 6 shows
the transmission delay of the conventional and proposed gateways when messages were transmitted from
the FlexRay to the CAN networks. In the tests, the brake node in the FlexRay network transmitted a
message (slot 30) to the vehicle node in the CAN network via the FlexRay-CAN gateway. Figure 6(a)
shows the average transmission delay that was obtained by transmitting the FlexRay message (slot 30) 100
times and then monitoring the CAN message transferred by the gateway to the CAN network. Figure 6(b)
shows the maximum transmission delay in the FlexRay-to-CAN message transmission repeated 100 times.
As shown in Figure 6(a), in both the conventional and proposed gateways, the average transmission delay
increased with increasing network traffic. Because increased network traffic reduces the chances of
transmitting the messages that are sent from the FlexRay to the CAN networks, the transmission delay was

Table II. Performance Using the HSS Index.

Conventional message-mapping-based gateway Proposed node-mapping-based gateway

N 962 285
V 9533 2324
D 326 97
E 3,109,282 225,611
 Performance Evaluation of Node-Mapping-Based Flexray-CAN Gateway for In-vehicle Networking System 259

(a) 600
conventional message-mapping-based gateway
proposed node-mapping-based gateway

transmission delay(µs)
400

200

0
30 42 52 62 72 82 86
network traffic(%)
(b) 6000
Downloaded by [Gazi University] at 06:18 16 April 2016

conventional message-mapping-based gateway

proposed node-mapping-based gateway
transmission delay(µs)

4000

2000

0
30 42 52 62 72 82 86
network traffic(%)

Figure 6. Performance of the FlexRay-CAN Gateway for FlexRay to CAN Transmission. (a) Average Transmission Delay from
FlexRay to CAN, (b) Maximum Transmission Delay from FlexRay to CAN.

increased. The network traffic was controlled by adjusting the message generation cycle of the node.
Figure 6(b) shows that the maximum transmission delay of the conventional and proposed gateways
increased with increasing network traffic. The reason behind such behavior is the same as for the average
transmission delay, described above. The maximum transmission delays were naturally higher than the
average transmission delays.
Figure 7 shows the transmission delay of the conventional and proposed gateways when messages
were transmitted from the CAN to the FlexRay networks. In the tests, the brake node in the CAN network
transmitted message 0x30 that was carried in a FlexRay static slot and sent to the vehicle node in the
FlexRay network. Figure 7(a) shows the average transmission delay in receiving the FlexRay message
containing CAN message 0x30, which was transmitted 100 times. Figure 7(b) shows the maximum
transmission delay in receiving the CAN message transmitted 100 times. The average transmission delay of
the conventional and proposed gateways maintained a certain constant level; irrespective of the increase in
network traffic (the graph is relatively flat). This is because once the message sent from the CAN network
passes through the gateway; it is transmitted in a scheduled FlexRay slot without any further delay in the
FlexRay network. Figure 7(b) shows that the maximum transmission delay of the conventional and
proposed gateways was also stable and not influenced by the increase in network traffic. The reason for this
result is the same as for the average transmission delay. The maximum transmission delays were higher
than the average transmission delays. This result demonstrates that the proposed and conventional
gateways have similar transmission delay characteristics.
 260 Intelligent Automation and Soft Computing

(a) 4000

transmission delay(µs) 2000

conventional message-mapping-based gateway

proposed node-mapping-based gateway
0
30 42 52 62 72 82 86
network traffic(%)
(b) 6000
Downloaded by [Gazi University] at 06:18 16 April 2016

transmission delay(µs)

4000

2000

conventional message-mapping-based gateway

proposed node-mapping-based gateway

0
30 42 52 62 72 82 86
network traffic(%)

Figure 7. Performance of FlexRay-CAN Gateway for CAN to FlexRay Transmission. (a) Average Transmission Delay, CAN to
FlexRay, (b) Maximum Transmission Delay, CAN to FlexRay.

SPSS version 14 was used to analyze the transmission delays of the FlexRay-CAN gateways via the
network traffic (SPSS 14: Quick Guide, 2007; SPSS 14.0 Developer’s Guide, 2005). Two independent
variables, gateway type and traffic, and their reciprocal action were analyzed. The statistical significance
level was set to a ¼ 0.05. Table III shows the statistical significance of transmission delay via network
traffic when messages are transmitted from the FlexRay to the CAN networks. There was a statistically
significant relationship when the network traffic was between 30% and 62%, and between 72% and 86%.

Table III. Comparison of P-value Results Between Transmission Delay and Network Traffic, FlexRay to CAN
(* p , 0.05, ** p , 0.01).

FlexRay network traffic

F –C 30% 42% 52% 62% 72% 82% 86%

30% ** ** **
CAN network traffic 42% * ** **
52% * **
62% ** **
72% ** *
82% ** ** *
86% ** ** ** **
 Performance Evaluation of Node-Mapping-Based Flexray-CAN Gateway for In-vehicle Networking System 261

Table IV. Comparison of P-value Results Between Transmission Delay and Network Traffic, CAN to FlexRay
(* p , 0.05, ** p , 0.01).

FlexRay network traffic

F –C 30% 42% 52% 62% 72% 82% 86%

CAN network traffic 30% *

42% *
52%
62%
72%
82%
86%
Downloaded by [Gazi University] at 06:18 16 April 2016

The transmission delay results showed a statistically significant effect for traffic (F(6, 1385) ¼ 13.83,
P ¼ 0.00), but not for gateway type (F(1, 1385) ¼ 5.63, P ¼ 0.18) nor gateway type £ traffic (F(6,
1385) ¼ 2.837, P ¼ 0.09). That is, the two gateways were not significantly different in terms of
transmission delay.
Table IV shows the statistical significance of transmission delay via network traffic when messages are
transmitted from the CAN to the FlexRay networks. There was a statistically significant relationship
only when the network traffic was 30% and 42%. This indicates that the transmission delay was not affected
by the increase in network traffic when messages are sent from the CAN to the FlexRay networks.
The transmission delay results showed a statistically significant effect for gateway type £ traffic
(F(6, 1399) ¼ 3.966, P ¼ 0.01), but not for traffic (F(6, 1399) ¼ 2.972, P ¼ 0.07) nor gateway type
(F(1, 1399) ¼ 0.49, P ¼ 0.824). This result shows that delay in transmitting messages from the CAN to the
FlexRay networks is not affected by the gateway type (message-mapping or node-mapping) and network
traffic conditions.
Our experimental results show that the proposed node-mapping-based FlexRay-CAN gateway yielded
a transmission delay level similar to the conventional message-mapping-based FlexRay-CAN gateway
while reducing the software complexity of the gateway, and was able to transmit messages even when the
message ID was changed. That is, the proposed gateway overcame the two drawbacks of the conventional
gateway.

4. SUMMARY AND CONCLUSIONS

We proposed a node-mapping-based FlexRay-CAN gateway to address the drawbacks of the conventional

message-mapping-based FlexRay-CAN gateway. We presented the structure and operation algorithm of
the proposed gateway and evaluated its feasibility using an experimental testbed. The conclusions derived
from this research are as follows:
The conventional gateway must change its software when a message is changed, and the gateway’s
software complexity increases rapidly as the number of messages to transfer increases. However, the
proposed gateway could overcome the drawbacks of the conventional gateway while maintaining a similar
transmission delay level.
The proposed gateway was evaluated in an experimental environment that simulated vehicle body and
chassis networks. The performance of the proposed gateway should be examined in realistic environments
where the FlexRay and CAN systems used in the real world are provided. In addition, it is important to
evaluate the performance of the messages that are transmitted via the proposed gateway at various cycle
 262 Intelligent Automation and Soft Computing

times. In the future, an integrated system that combines the CAN and FlexRay networks will be
implemented using the proposed gateway and its performance will be investigated.

ACKNOWLEDGEMENTS
This research was partially supported by the MSIP (The Ministry of Science, ICT and Future Planning), Korea, under the CITRC
(Convergence Information Technology Research Center) support program (NIPA-2014-H0401-14-1001) supervised by the NIPA
(National IT Industry Promotion Agency).

REFERENCES
SPSS 14.0 Developer’s Guide. SPSS Inc (2005).
SPSS 14: Quick Guide. (2007). Leeds Metropolitan University.
Downloaded by [Gazi University] at 06:18 16 April 2016

Baronti, F., Petri, E., Saponara, S., Fanucci, L., Roncella, R., Saletti, R., D’Abramo, P., & Serventi, R. (2011). Design and verification
of hardware building blocks for high-speed and fault-tolerant in-vehicle networks. IEEE Transactions on Industrial Electronics,
58, 792–801.
Chen, X., & Zhang, G. (2011). Design of automotive gateway based on FlexRay. International Conference on Information and Control
Engineering (ICEICE), 4897–4900.
Dong, Z. H., Piao, Z. Y., Jang, I. G., Chung, J. G., & Lee, C. D. (2012). Design of FlexRay-MOST gateway using static segments and
control messages. IEEE International Symposium on Circuits and Systems (ISCAS), 536–539.
Freescale Semiconductor, Inc. (2010). MC9S12XF512 Reference Manual.
Han, SC, & Han, Y. H. (2014). IT convergence with traditional industries and short-term research and development strategy in Korea.
Intelligent Automation & Soft Computing, 20, 3–14.
Kelkar, S., & Kamal, R. (2014). Adaptive fault diagnosis algorithm for controller area network. IEEE Transactions on Industrial
Electronics, 61, 5527–5537.
Lee, M. J., Oh, J. Y., & Kang, S. J. (2011). Design of multimedia stream channel arbiter in home network gateway. IEEE Transactions
on Consumer Electronics, 57, 1661–1669.
Li, H., Zhang, H., Xia, F., & Huang, W. (2009). Research on communication controller between FlexRay and Modbus. International
Conference on ICSAP, 73–76.
Ouedraogo, L., & Kumar, R. (2014). Computation of the precise worst-case response time of FlexRay dynamic messages. IEEE
Transactions on Automation Science and Engineering, 11, 537 –548.
Park, I., & Sunwoo, M. (2011). FlexRay network parameter optimization method for automotive applications. IEEE Transactions on
Industrial Electronics, 58, 1449–1459.
Qin, W., Feng, K., & Xiuwei, F. (2010). Research of automotive network FlexRay/CAN gateway based on DSP. International
Conference on Future Computer and Communication, V2-837–V2-840.
Schmidt, E. G., Alkan, M., Schmidt, K., Yuruklu, E., & Karakaya, U. (2010). Performance evaluation of FlexRay/CAN networks
interconnected by a gateway. International Symposium on Industrial Embedded Systems, 209 –212.
Schmidt, KW (2014). Robust priority assignments for extending existing controller area network applications. IEEE Transactions on
Industrial Informatics, 10, 578 –585.
Seo, H. S., Kim, B. C., Park, P. S., Lee, C. D., & Lee, S. S. (2013). Design and implementation of a UPNP-CAN gateway for
automotive environments. International Journal of Automotive Technology, 14, 91 –99.
Seo, S. H., Ki, J. H., Hwang, S. H., Kwon, K. H., & Jeon, J. W. (2009). An evaluation of the FlexRay-CAN gateway-embedded system
in the HEV test bench. IEEE International Symposium on Industrial Electronics, 664–669.
Sheng, Y., & Shijie, Z. (2010). A survey on metrics of software complexity. 2nd IEEE International Conference on Information
Management and Engineering, 352– 356.
Taylor, J. E., Amor-Segan, M. M., Dhadyalla, G. G., & Jones, R. P. (2014). Discerning the operational state of a vehicle’s distributed
electronic systems from vehicle network traffic for use as a fault detection and diagnosis tool. International Journal of Automotive
Technology, 15, 441 –449.
Tindell, K., & Burns, A. (1994). Guaranteeing message latencies on controller area network (CAN). Proceedings of the first
international CAN conference. 1.2–1.11.
Tindell, K., Burns, A., & Wellings, A. J. (1995). Calculating controller area network (CAN) message response times. Control
Engineering Practice, 3, 1163–1169.
Tu, H., Sun, W., & Zhang, Y. (2009). The research on software metrics and software complexity metrics. International Forum on
Computer Science-Technology and Applications, 131 –136.
Vector, Inc. (2010). CANoe.FlexRay 7.2 Product Catalog.
Zhao, R., Qin, G., & Liu, J. (2010). Gateway system for CAN and FlexRay in automotive ECU networks. International Conference on
Information Networking and Automation (ICINA), 49–53.
 Performance Evaluation of Node-Mapping-Based Flexray-CAN Gateway for In-vehicle Networking System 263

NOTES ON CONTRIBUTORS
Man Ho Kim received Ph. D. degree from Pusan National University, Busan, Korea, in 2008. He is a
research professor in the Leader in Industry-university Cooperation, Pusan National University, Busan,
Korea. His research interests include fault-tolerant system with hardware redundancy structure, in-vehicle
networking system, embedded microcontroller system design, and driving behavior and workload
analysis of driving vehicle.

Suk Lee received Ph. D. degree from The Pennsylvania State University, University Park, in 1990. He is a
professor in the School of Mechanical Engineering, Pusan National University, Busan, Korea. His
Downloaded by [Gazi University] at 06:18 16 April 2016

research interests are industrial network, in-vehicle network, and home network.

Kyung Chang Lee received Ph.D. degree from Pusan National University, Busan, Korea, in 2003. He is a
professor in the Department of Control and Instrumentation Engineering, Pukyong National University,
Busan, Korea. His research interests are embedded network system, industrial network, robotic network,
in-vehicle network, home network, wireless sensor network, and networked control system.