JP7122871B2 - VEHICLE GATEWAY DEVICE, METHOD AND PROGRAM - Google Patents

Description

The present invention relates to an in-vehicle gateway device, method and program.

In recent years, automobiles are equipped with a large number of devices called electronic control units (ECUs). These ECUs are connected to each other by a network to send and receive messages. In order to transmit and receive messages between ECUs belonging to different networks, a relay device (also referred to as an in-vehicle gateway device or simply a gateway) that connects networks and relays messages is used. 2. Description of the Related Art In recent years, the number of ECUs, the number of messages transmitted and received between ECUs, and the number of messages relayed by gateways have increased rapidly as automobiles have become more sophisticated and sophisticated.

Many technologies related to gateways have been proposed. For example, Patent Literature 1 discloses a technique in which a gateway having a transmission buffer and a reception buffer transmits unsent data buffered in the transmission buffer in order of priority.

Japanese Unexamined Patent Application Publication No. 2017-212728

Upon receiving a message, the gateway performs routing processing to determine the destination of the message and determines which messages to send and in what order. The gateway allocates computational resources to these operations and temporarily stores the message until its transmission is complete. If a new message arrives before an existing message has finished processing, a buffer overrun can occur and the newly arrived message can be lost.

Message loss may be acceptable in consumer applications. For example, the Internet's TCP/IP protocol tolerates message loss because of message retransmission. On the other hand, real-time systems, such as in-vehicle systems, cannot tolerate message loss. This is because excessive delay associated with retransmission is not allowed.

A simple solution to avoid message loss would be to equip the gateway with a sufficiently powerful computer (eg CPU). However, considering that the number and size of messages may increase further in the future, a solution that simply includes a powerful computer is not desirable.

Moreover, even if sufficiently powerful computers were employed to avoid message loss, there would still be the possibility of delays in sending messages. This is because, under conditions of high network load, the gateway may prioritize receiving messages over sending and accumulate received messages without sending them. Message delay is not acceptable in a system such as an in-vehicle system that requires real-time performance. While the above Patent Document 1 discloses that unsent data is transmitted in order of priority, it does not disclose any device for avoiding message loss.

SUMMARY OF THE INVENTION Accordingly, the present invention has been made in view of the above problems. It is another object of the present invention to provide an in-vehicle gateway device, method and program.

In order to solve the above problems, according to one aspect of the present invention, there is provided an in-vehicle gateway device (100) connected to N networks and relaying messages between different networks, comprising: N first storage units (120) for storing messages awaiting routing processing among messages received from each; a second storage unit (130), selecting one message as a processing object from the messages waiting for the routing processing and the messages waiting for the transmission processing, and executing the routing processing or the transmission processing of the message to be processed; and a controller (150) for allocating computational resources.

In order to solve the above problems, according to another aspect of the present invention, there is provided a method executed by an in-vehicle gateway device (100) connected to N networks and relaying messages between different networks. a step (S104 or S306) of storing messages awaiting routing processing among messages received from each of the N networks in N first storage units (120); a step (S212, S214, S412, or S414) of storing information indicating messages awaiting transmission processing in one or more second storage units (130); and said messages awaiting routing processing and said messages awaiting transmission processing. A step of selecting one message from among the messages to be processed and allocating computational resources to the routing process or the transmission process of the message to be processed (S208, S218, S222, S408, S434 or S436, and S212, S214, S224, S412, S414, or S438).

In order to solve the above problems, according to another aspect of the present invention, a computer that controls an in-vehicle gateway device (100) that is connected to N networks and relays messages between different networks includes N a step (S104 or S306) of storing messages awaiting routing processing among messages received from each of said networks in N first storage units (120); a step (S212, S214, S412, or S414) of storing in one or more second storage units (130) information indicating messages of; A step of selecting one message to be processed and allocating computational resources to the routing process or the transmission process of the message to be processed (S208, S218, S222, S408, S434 or S436, and S212, S214, S224, S412 , S414, or S438) are provided.

As described above, according to the present invention, there is provided a new and improved in-vehicle gateway device, method, and program capable of both avoiding incoming message loss and securing transmission opportunities.

It is a figure which shows an example of a structure of the gateway based on one Embodiment of this invention. FIG. 7 is a diagram for explaining an example of message reception processing by the gateway according to the first embodiment; FIG. 4 is a diagram for explaining an example of message routing processing by a gateway according to the embodiment; It is a figure for demonstrating an example of the transmission processing of the message by the gateway which concerns on the same embodiment. It is a flowchart which shows an example of the flow of the reception process performed in the gateway which concerns on the same embodiment. 4 is a flowchart showing an example of the flow of prioritization processing, routing processing, and transmission processing executed in the gateway according to the same embodiment; FIG. 11 is a diagram for explaining an example of message reception processing by a gateway according to the second embodiment; FIG. 4 is a diagram for explaining an example of message routing processing by a gateway according to the embodiment; It is a figure for demonstrating an example of the transmission processing of the message by the gateway which concerns on the same embodiment. It is a flowchart which shows an example of the flow of the reception process performed in the gateway which concerns on the same embodiment. It is a flowchart which shows an example of the flow of a prioritization process, a routing process, and a transmission process which are performed in the gateway which concerns on the same embodiment. It is a flowchart which shows an example of the flow of a prioritization process, a routing process, and a transmission process which are performed in the gateway which concerns on the same embodiment. FIG. 10 is a diagram illustrating an example of a configuration of a gateway according to a comparative example;

Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. In the present specification and drawings, constituent elements having substantially the same functional configuration are denoted by the same reference numerals, thereby omitting redundant description.

<<1. Configuration example >>
FIG. 1 is a diagram showing an example of the configuration of a gateway 100 according to one embodiment of the invention. The gateway 100 is an in-vehicle gateway device that is connected to multiple networks and relays messages between different networks. As shown in FIG. 1, the gateway 100 includes a network interface 110, N first storage units 120 (queues #1 to #N), and one or more second storage units 130 (queues #N+1 to #M). ), a third storage unit 140 and a control unit 150 .

(1) Network interface 110
The network interface 110 is an interface that mediates communication between the gateway 100 and other devices. The network interface 110 is connected to any in-vehicle network such as CAN (Controller Area Network) or CAN-FD (CAN with Flexible Data rate). Network interface 110 has a plurality of channels, each channel connected to a different network. In the example shown in FIG. 1, network interface 110 has N channels, channels #1 through #N. In addition, the number attached after "#" in FIG. 1 is an index. That is, the gateway 100 is connected to N networks. Note that N is an integer of 2 or more.

The network interface 110 has N receive buffers 111 (111#1 to 111#N) (corresponding to a fourth storage unit) that store messages received from each of the N networks. The reception buffer 111 is associated one-to-one with a channel (eg, CAN channel) and stores messages (eg, CAN message) that have arrived at the associated channel. For example, receive buffer 111#1 stores messages arriving on channel #1.

The network interface 110 has N transmission buffers 112#1 to 112#N (corresponding to fifth storage units) for storing messages to be transmitted to each of the N networks. The transmission buffers 112#1 to 112#N are also collectively referred to as the transmission buffers 112. FIG. The transmission buffer 112 is associated with a channel on a one-to-one basis and stores messages transmitted from the associated channel. For example, transmission buffer 112#1 stores messages transmitted from channel #1.

(2) First storage unit 120
The N first storage units 120 store messages awaiting routing processing among messages received from each of the N networks. The routing process is a process of checking a received message against a routing table to determine a destination network (that is, a channel). The routing process may further include various processes such as security-related filtering based on message headers and payloads. The first storage unit 120 can be implemented as a queue, for example. The first storage unit 120 is hereinafter also referred to as a routing queue 120 . In FIG. 1, queues #1 to #N correspond to the routing process waiting queue 120. FIG.

(3) Second storage unit 130
One or more second storage units 130 store information indicating messages that have undergone routing processing and are waiting for transmission processing. The transmission process includes the process of transmitting the message after the routing process to the determined destination network. The second storage unit 130 can be implemented as a queue, for example. The second storage unit 130 is hereinafter also referred to as a transmission processing waiting queue 130 . In FIG. 1, queues #N+1 to #M correspond to the routing process waiting queue 120. FIG. Note that M≧N+1.

Here, "information indicating a message waiting for transmission processing" is information indicating a pointer to a message waiting for transmission processing or a storage unit (a third storage unit 140 described later) that stores a message waiting for transmission processing. , for example pointers. The transmission processing waiting queue 130 may store the messages themselves that are waiting for transmission processing.

(4) Third storage unit 140
The third storage unit 140 stores messages that have undergone routing processing and are waiting for transmission processing. The third storage unit 140 is implemented by any data structure for storing messages. The third storage unit 140 includes at least one of an MPFO (Most Priority First Out) list 141 or a FIFO (First In First Out) list 142 . In the example shown in FIG. 1, the third storage unit 140 contains both the MPFO list 141 and the FIFO list 142. FIG. An MPFO list is a data structure in which stored messages are retrieved in order of highest priority. A FIFO list is a data structure in which previously stored messages are retrieved in order. For example, CAN control messages are stored in the MPFO list 141 and transmitted in order of priority. CAN diagnostic messages, on the other hand, are stored in a FIFO list 142 and sent on a first-come, first-served basis.

(5) Control unit 150
The control unit 150 functions as an arithmetic processing device and a control device, and controls the overall operation of the gateway 100 according to various programs. The control unit 150 is implemented by a CPU, microprocessor, electric circuit, DSP and/or ASIC.

The control unit 150 receives messages arriving at the gateway 100 and performs various processes for forwarding the messages. Specifically, the control unit 150 performs reception processing, prioritization processing, routing processing, and transmission processing. The control unit 150 allocates computational resources (for example, CPU processing time) to these processes. That is, the control unit 150 selects a process for allocating computational resources from these processes, and executes the selected process. Regarding the routing process and the transmission process, the control unit 150 selects one message from among the messages waiting for the routing process and the messages waiting for the transmission process as a processing target, and calculates the routing process or the transmission process for the message to be processed. Allocate resources. The selection of messages for processing is made according to the results of the prioritization process.

An example of the configuration of the gateway 100 common to the first embodiment and the second embodiment described below has been described above.

<<2. First Embodiment>>
In this embodiment, a round-robin method is adopted for the prioritization process. Specifically, the gateway 100 selects messages corresponding to queues selected by a round-robin strategy from the queue 120 waiting for routing processing and the queue 130 waiting for transmission processing, to be processed. Since the queues are selected in order in a round-robin fashion and allocated computational resources, routing and transmission processes are equally allocated computational resources. As a result, it is possible to achieve both avoidance of incoming message loss and securing transmission opportunities.

<2.1. Various processing>
Various processes executed by the gateway 100 according to this embodiment will be described in detail below. The configuration of the gateway 100 according to this embodiment is as described above with reference to FIG. In the following description, it is assumed that N=6 and M=8. In this case, 25% ((M−N)/M) of the computational resources available for routing and transmission processing are allocated for transmission processing, and the remaining 75% (N/M)) is allocated for routing processing. be done.

(1) Reception Processing Reception processing is processing for moving messages stored in the reception buffer 111 to the routing processing waiting queue 120 . Note that moving a message is a concept that includes copying a message to the destination and deleting the message from the source. The control unit 150 allocates computational resources to the reception process prior to the prioritization process, the routing process, and the transmission process. This makes it possible to avoid loss of incoming messages as much as possible.

When a message arrives from the network and is stored in receive buffer 111 of network interface 110, an interrupt is triggered. The interrupt moves the received message to the end of the routing queue 120 corresponding to the channel through which the message arrived. An example of the reception process will be described below with reference to FIG.

FIG. 2 is a diagram for explaining an example of message reception processing by the gateway 100 according to this embodiment. FIG. 2 illustrates the data structure and data flow within gateway 100 . Queues #1 to #6 are queues 120 waiting for routing processing, and queues #7 and #8 are queues 130 waiting for transmission processing. Queue #1 stores message 10-1, queue #2 stores message 10-2, and queue #6 stores message 10-3. Also, the MPFO list 141 stores the message 10-4, and the FIFO list 142 stores the message 10-5.

In the example shown in FIG. 2, message 10-6 arriving on channel #1 is stored in receive buffer 111#1 of channel #1. Message 10-6 is then moved to the end of queue #1 and deleted from receive buffer 111#1 of channel #1.

In each queue and list, the lower the message is, the earlier the message is output, and the message at the bottom is the first message. For example, the head of queue #1 is message 10-1 and the second is message 10-6.

(2) Prioritization process Prioritization process is to select a queue to be processed. Specifically, the control unit 150 selects one queue from the routing processing waiting queue 120 and the transmission processing waiting queue 130 as a queue to be processed. Processing corresponding to the queue selected in the prioritization process is then performed. That is, if the selected queue is the routing process waiting queue 120, the routing process is performed, and if the selected queue is the transmission process waiting queue 130, the transmission process is performed.

In this embodiment, the prioritization process is performed according to a round-robin scheme. Specifically, the control unit 150 selects one queue from the non-empty routing processing waiting queue 120 and transmission processing waiting queue 130 as a processing target queue in a round-robin manner. That is, non-empty queues are selected in order as queues to be processed. Here, non-empty queues are the routing processing waiting queue 120 storing one or more messages, and the transmission processing waiting queue 130 when messages are stored in the MPFO list 141 or FIFO list 142 . If MPFO list 141 or FIFO list 142 is not empty, transmission queue 130 may store pointers to messages stored in MPFO list 141 or FIFO list 142 . Then, whether or not the transmission processing waiting queue 130 is empty may be determined based on the presence or absence of the pointer. In the following, non-empty queues are also referred to as active queues. In the example shown in FIG. 2, if queue #5 was selected last time, queue #6 is selected next time.

The active queues are selected for processing in a round-robin fashion in order to prevent excessive computational resources being allocated to either the routing or transmission processing. Therefore, it is possible to fairly distribute computational resources to routing processing and transmission processing. N and M can be set arbitrarily. The balance of N and M determines the balance of computing resources distributed between routing and transmission processing.

As a technique related to such prioritization processing, Reference [2] discloses a packet switch that extracts packets from queues selected by a round-robin method and transmits the packets. However, the object of distribution in this document is channel bandwidth, not computational resources as in the present invention.

(3) Routing Processing The control unit 150 selects the message corresponding to the processing target queue selected by the prioritization processing as the processing target message. More specifically, when the queue to be processed selected by the prioritization process is the routing process waiting queue 120, the control unit 150 selects the routing process waiting queue 120 at the head of the selected routing process waiting queue 120 (that is, the first stored queue). ) message as the message to be processed. Then, the control unit 150 allocates computational resources to the routing process of the message to be processed.

The routing process involves matching the message to a routing table to determine the channel to send it to. The routing process also includes various checks such as security checks of messages by firewalls. The routing process also includes moving messages from the routing process waiting queue 120 to the MPFO list 141 or FIFO 142 list. The control unit 150 determines the data structure of the destination according to the content of the message. For example, the control unit 150 moves CAN control messages to the MPFO list 141 and CAN diagnostic messages to the FIFO list 142 . In addition, the control unit 150 can store pointers to messages stored in the MPFO list 141 or the FIFO list 142 in the transmission queue 130 .

An example of routing processing will be described below with reference to FIG. FIG. 3 is a diagram for explaining an example of message routing processing by the gateway 100 according to the present embodiment. FIG. 3 shows the data structure in the gateway 100 and the data flow following FIG. In the example shown in FIG. 3, queue #6 is assumed to be selected by the prioritization process. In this case, the routing process for message 10-3 at the head of queue #6 is executed. Message 10-3 is thereby stored in the associated MPFO list 141 and deleted from queue #6. Assume that message 10-3 has a lower priority than existing message 10-4 in MPFO list 141. The storage of CAN messages in MPFO lists or FIFO lists is disclosed in detail in reference [1].

(4) Transmission Processing The control unit 150 selects the message corresponding to the processing target queue selected by the prioritization processing as the processing target message. Specifically, when the processing target queue selected by the prioritization process is the transmission processing waiting queue 130, the control unit 150 selects the message corresponding to the selected transmission processing waiting queue 130 as the processing target message. do. Then, the control unit 150 allocates computational resources to the transmission process of the message to be processed.

Here, the messages corresponding to the queue 130 waiting for transmission processing are messages stored in the third storage unit 140 . Specifically, when the MPFO list 141 is not empty, the control unit 150 selects the message with the highest priority in the MPFO list 141 as the message to be processed. On the other hand, when the MPFO list 141 is empty, the control unit 150 selects the first message in the FIFO list 142 as the message to be processed. Note that if both the MPFO list 141 and the FIFO list 142 are empty, subsequent transmission processing is not executed.

The selected message to be processed is stored in the transmission buffer 112 corresponding to the destination channel determined by the routing process, and deleted from the MPFO list 141 or FIFO list 142 . The messages stored in transmission buffer 112 are then transmitted.

An example of the transmission process will be described below with reference to FIG. FIG. 4 is a diagram for explaining an example of message transmission processing by the gateway 100 according to this embodiment. FIG. 4 shows the data structure in gateway 100 and the flow of data following FIG. In the example shown in FIG. 4, queue #7 is assumed to be selected by the prioritization process. In this case, since the MPFO list 141 is not empty, the message 10-4 with the highest priority among those stored in the MPFO list 141 is stored in the transmission buffer 112#5 of the channel #5, which is the transmission destination, and the channel Sent from #5. Also, the transmitted message 10-4 is deleted from the MPFO list 141. FIG.

(5) Supplement Prioritization processing, routing processing, and transmission processing can be executed as interrupts or as tasks, but in order to achieve low latency, they are preferably executed as interrupts. This is because interrupts are executed with priority over tasks. However, interrupts for receive processing take precedence over interrupts and tasks for prioritization, routing and transmit processing.

Triggers for prioritization, routing or sending processes can be set using task or interrupt flags. The flag indicates, for example, whether or not there is a non-empty queue or list in the routing queue 120, transmission queue 130, MPFO list 141, and FIFO list 142. In this case, the receive processing interrupt checks the flag and, if the non-empty flag is set, activates the prioritization processing, routing processing, or transmit processing interrupt or task. Furthermore, the interrupts or tasks of the prioritization processing, the routing processing, or the transmission processing are placed in the routing processing waiting queue 120, the transmission processing waiting queue 130, the MPFO list 141, and the FIFO list after executing the prioritization processing, routing processing, or transmission processing. 142 is empty, and if not, call itself again.

<2.2. Process Flow>
(1) Flow of Reception Processing FIG. 5 is a flowchart showing an example of the flow of reception processing executed in the gateway 100 according to this embodiment. As shown in FIG. 5, when a message arrives on channel #X, the controller 150 buffers the incoming message in the receive buffer 111#X corresponding to channel #X (step S102). Next, the control unit 150 moves the message buffered in the reception buffer 111#X to the routing process waiting queue 120 (queue #X) corresponding to the channel #X (step S104). Then, the control unit 150 triggers prioritization processing, routing processing, or transmission processing (step S106).

(2) Flow of Prioritization Processing, Routing Processing, and Transmission Processing FIG. 6 is a flowchart showing an example of the flow of prioritization processing, routing processing, and transmission processing executed by the gateway 100 according to the present embodiment. As shown in FIG. 6, first, when the prioritization process, the routing process, and the transmission process are triggered by the reception process (step S202), the control unit 150 , selects a non-empty queue next to the previously selected queue (step S204). Let X be the index of the queue to be selected. Next, the control unit 150 determines whether the selected queue #X is the routing process waiting queue 120 or the transmission process waiting queue 130 (step S206). If the selected queue #X is the routing process waiting queue 120, the process proceeds to step S208. On the other hand, if the selected queue #X is the transmission process waiting queue 130, the process proceeds to step S226. Note that these processes correspond to prioritization processes.

In step S208, control unit 150 selects the top message from queue #X. Next, the control unit 150 determines whether the selected message is an MPFO message or a FIFO message (step S210). If the selected message is a FIFO message, the control unit 150 performs routing and moves the message to the FIFO list 142 (step S212). On the other hand, if the selected message is an MPFO message, the control unit 150 performs routing and moves the message to the MPFO list 141 (step S214). Here, pointers to messages stored in MPFO list 141 or FIFO list 142 can be stored in queue 130 waiting to be processed. After step S212 or S214, the process proceeds to step S216. Note that these processes correspond to routing processes.

In step S216, control unit 150 determines whether MPFO list 141 is empty. If it is determined that the MPFO list 141 is not empty, the control unit 150 selects the highest priority message from the MPFO list 141 (step S218). After that, the process proceeds to step S224. On the other hand, when it is determined that the MPFO list is empty, the control unit 150 determines whether the FIFO list 142 is empty (step S220). If the FIFO list 142 is determined to be empty, the process returns to step S204 again. On the other hand, if it is determined that FIFO list 142 is not empty, controller 150 selects the first message from FIFO list 142 (step S222). After that, the process proceeds to step S224. In step S<b>224 , the control unit 150 transmits the message to the destination channel determined by the routing process, and deletes the transmitted message from the MPFO list 141 or FIFO list 142 . Along with this, the pointer to the deleted transmitted message is deleted from the queue 130 waiting for transmission processing. After that, the process proceeds to step S226. Note that these processes correspond to the transmission process.

In step S226, control unit 150 determines whether or not the queue, MPFO list 141 and FIFO list 142 are empty. If it is determined that at least one is not empty, the process returns to step S204. On the other hand, if it is determined that all are empty, the process ends.

<<3. Second Embodiment>>
<3.1. Overview>
In the first embodiment, non-empty queues were selected in order in a round-robin fashion to allocate computing resources. Such an allocation scheme is not completely fair for two reasons.

The first reason is that the processing target is selected regardless of the length of the message in the round-robin method, although the processing time for long messages is longer than the processing time for short messages. This means that long messages can get more computational resources compared to short messages, in other words long messages have priority over short messages.

The second reason is that although the processing time for routing processing is longer than the processing time for transmission processing, in the round robin method, processing targets are selected regardless of the difference in processing time. This means that the routing process can acquire more computational resources than the transmission process, in other words, the routing process has priority over the transmission process.

Therefore, in the second embodiment, a mechanism is provided that can establish fairness in allocation of computational resources even when a processing time occurs depending on the size of the message and the type of processing applied. do. Specifically, the control unit 150 allocates computational resources according to the message processing time.

Here, it is considered ideal that the granularity of sharing the computational resource (for example, the processing time of the CPU) is the time of one clock. In other words, it can be said that it is ideal to process the message at the top of the active queue while switching the queue to be processed every clock cycle, that is, to process in parallel. However, such an ideal mechanism is not realistic. This is because at least the routing process cannot be completed in one clock cycle and is difficult to divide (ie, it is an atomic operation). The process of sending a message is also an atomic operation. Interrupting the routing process and returning after several clocks, or interrupting message transmission and returning after several clocks, is not realistic.

Therefore, in the priority ranking process, the control unit 150 according to the present embodiment selects queues to be processed based on the scheduled time at which message processing ends (hereinafter also referred to as the scheduled end time). Specifically, the control unit 150 calculates the scheduled end time of each process of the message waiting for routing processing and the message waiting for transmission processing, and selects the message with the earliest scheduled end time as the message to be processed. When a new message i (or a pointer to message i) is stored in queue #X, control unit 150 takes into consideration the scheduled end time of messages stored in queue #X before message i, Calculate the expected end time of message i. Specifically, the control unit 150 determines whether the scheduled end time of a message processed before the message whose scheduled end time is to be calculated is stored in the same queue as the message whose scheduled end time is to be calculated, or the message whose scheduled end time is to be calculated is queued. By adding the processing time of the message to be calculated to the later one of the times stored in , the estimated end time of the message to be calculated is calculated. Expressed as a formula, the control unit 150 calculates the expected end time F _X [i] according to the following formula.
F _X [i]=Max(F _X [i−1], A _X [i])+P _X [i] (1)

where X is the queue index and i is the message index in queue #X. F _X [i−1] is the scheduled end time of the message immediately before message i stored in queue #X. A _X [i] is the time when message i was stored in queue #X. P _X [i] is the message processing time. However, the processing time depends on the size of the message to be calculated and the type of processing applied. That is, P _X [i] has a larger value as the size of the message is larger, and a smaller value as the size of the message is smaller. Also, P _X [i] indicates whether the processing applied to the message is routing processing or transmission processing (that is, whether queue #X is the routing processing waiting queue 120 or the transmission processing waiting queue 130). determined according to Note that P _X [i] can be obtained by referring to a lookup table stored in advance in a storage unit (not shown). Such a lookup table is a table containing a plurality of correspondences between message size and type of processing to be applied and processing time.

If queue #X is empty before message i is stored (i.e., i=1), control unit 150 calculates expected end time F _X [1] according to the following equation.

F _X [1] = A _X [1] + P _X [1] (2)

The control unit 150 calculates F _X [i] for the messages stored in each of the active queues according to equation (1) or (2) above. Then, the control unit 150 selects the message with the earliest scheduled end time F _X [i] among the messages at the top of each queue as the message to be processed.

By selecting messages to be processed based on the scheduled end time, the gateway 100 can process the messages in the same manner as the ideal mechanism described above, as described below.

Specifically, since P _X [i] depends on the size of the message, long and short messages can be fairly allocated computational resources. Also, since P _X [i] depends on the type of processing applied to the message, it is possible to fairly allocate computational resources to the routing and transmission processes. Thus, in the second embodiment, the unfairness of the allocation method due to the two reasons described above can be resolved.

Also, since F _X [i−1] or A _X [i] is added to the calculation of F _X [i], the scheduled end time is calculated based on the scheduled start time of processing message i. . This makes it possible to calculate an estimated end time of a message that reflects the pace of message processing, which may vary from queue to queue.

As described above, in this embodiment, the scheduled end time is calculated by adding the message processing time to the scheduled start time of the process. Therefore, in this embodiment, the order in which message processing ends is the same as the order in which message processing ends in the ideal mechanism described above. In other words, selecting a message to be processed based on the estimated end time models the message processing while shifting the processing target from one queue to another queue at each clock cycle in the ideal mechanism described above. It can be said that there is In fact, it is mathematically proven in prior literature [3] that essentially the same behavior as the ideal mechanism described above is achieved when selecting messages for processing based on expected end time. .

However, when messages at the top of active queues are processed in parallel as in the ideal mechanism described above, the message processing speed changes depending on the number of active queues. To reflect this concept in the prioritization process, virtual time is introduced in the present embodiment. The virtual time is updated accordingly. The virtual time is obtained by dividing the difference between the current real time and the real time of the previous update by the number of non-empty queues in the routing process waiting queue 120 and the transmission process waiting queue 130, and the virtual time of the previous update. is updated by adding to Expressing it numerically, the control unit 150 calculates the virtual time Tvir according to the following equation.
Tvir=Tvirold+(Tnow-Told)/Q (3)

Here, Tvirold is the virtual time of the previous update. Tnow is the current real time. Told is the real time when the virtual time was last updated. Q is the number of active queues. Referring to Equation (3) above, the update width of the virtual time can increase or decrease according to the number of active queues. For example, if there is one active queue, the virtual time is the same as the real time. On the other hand, if there are two active queues, virtual time advances twice as slow (ie, half as fast) as compared to real time. This corresponds to an increase or decrease in the processing speed of messages processed in parallel due to an increase or decrease in the number of active queues in the ideal scheme described above.

The virtual time is updated each time a new message (or a new pointer) is added to the routing queue 120 or transmission queue 130 . Alternatively, the virtual time may be updated each time the number of non-empty, ie active queues in the routing queue 120 and transmission queue 130 increases or decreases.

The virtual time is used instead of the time A _X [i] at which the message to be calculated was stored in the queue in calculating the expected end time. That is, the control unit 150 sets the calculation target message to the scheduled end time or the virtual time of the message that is stored in the same queue as the calculation target message and is processed before the calculation target message, whichever is later. By adding the processing time of the message, the estimated end time of the message to be calculated is calculated. Expressed as a formula, the control unit 150 calculates the expected end time F _X [i] according to the following formula.
F _X [i]=Max(F _X [i−1], Tvir)+P _X [i] (4)

If queue #X is empty before message i is stored (i.e., i=1), control unit 150 calculates expected end time F _X [1] according to the following equation.

F _X [1]=Tvir+P _X [1] (5)

By introducing virtual time, changes in message processing speed caused by changes in the number of active queues can be reflected in the scheduled end time. This allows the gateway 100 to achieve allocation of computational resources that is closer to the ideal mechanism described above.

<3.2. Various processing>
Various processes executed by the gateway 100 according to this embodiment will be described in detail below. The configuration of the gateway 100 according to this embodiment is as described above with reference to FIG. In the following description, it is assumed that N=6 and M=7. In this case, approximately 14% ((M−N)/M) of the computational resources available for routing and transmission processing are allocated for transmission processing, and the remaining approximately 86% (N/M) for routing processing. assigned to.

(1) Receive Processing When a message arrives from the network and is stored in the receive buffer 111 of the network interface 110, an interrupt is triggered. The interrupt moves the received message to the end of the routing queue 120 corresponding to the channel through which the message arrived. At that time, first, the control unit 150 calculates the virtual time Tvir based on the above formula (3).

Next, the control unit 150 calculates the estimated end time F _X [i] based on the above formula (4) or (5), and stores it in the routing queue 120 in association with the message. Q is incremented if the routing process queue 120 at the destination of the message is empty before the message is moved. Also, the control unit 150 overwrites Tvirold and Told with Tvir and Tnow as in the following equations.
Tvirold=Tvir, Told=Tnow (6)

An example of the reception process will be described below with reference to FIG.

FIG. 7 is a diagram for explaining an example of message reception processing by the gateway 100 according to this embodiment. FIG. 7 illustrates the data structure and data flow within the gateway 100 . Queues #1 to #6 are queues 120 waiting for routing processing, and queue #7 is queue 130 waiting for transmission processing. Queue #1, Queue #2, and Queue #4 contain messages 10-1, 10-2, and 10-3, and estimated end times F ₁ [1], F ₂ [1], F of each message. ₄ [1] is stored. Also, the MPFO list 141 stores the message 10-4, and the FIFO list 142 stores the message 10-5. Queue #7 also contains a pointer 11-1 to message 10-4, an expected end time F ₇ [1] of message 10-4, a pointer 11-2 to message 10-5, and a message 10-5. Estimated end time F ₇ [2] is stored.

In the example shown in FIG. 7, the message 10-6 arriving on channel #1 is stored in the receive buffer 111#1 of channel #1, then placed at the end of queue #1, and is placed in the receive buffer of channel #1. removed from Since message 10-1 is already stored at the head of queue #1 and is not empty, message 10-6 is stored second in queue #1 and Q is not incremented.

In each queue and list, the lower the message is, the earlier the message is output, and the message at the bottom is the first message. For example, the head of queue #1 is message 10-1 and the second is message 10-6.

(2) Prioritization process Prioritization process is to select a queue to be processed. Specifically, the control unit 150 selects one queue from the routing processing waiting queue 120 and the transmission processing waiting queue 130 as a queue to be processed. In this embodiment, the queue storing the message with the earliest end time F _X [i] among the top messages of each queue is selected as the queue to be processed.

(3) Routing Processing and Transmission Processing After the prioritization processing, processing corresponding to the selected queue is executed. The routing processing and transmission processing performed here are the same as those in the first embodiment. Differences will be described below.

- First difference After the routing process, the message is stored in the transmission process waiting queue 130 . At this time, the control unit 150 calculates the virtual time Tvir based on the above equation (3), and overwrites Tvirold and Told with Tvir and Tnow based on the above equation (6). Also, the control unit 150 calculates the expected end time F _X [i] based on the above formula (4) or (5), and stores it in the transmission processing waiting queue 130 in association with the message.

However, when a routed message is stored in the MPFO list 141 or FIFO list 142, the message does not have to be copied to the queue 130 waiting for transmission processing. In the following description, it is assumed that F _X [i] and a pointer to a message stored in the MPFO list 141 or FIFO list 142 are stored in the transmission queue 130 .

Here, in the MPFO list 141, the transmission order of messages is determined based on the priority of each message. Therefore, when new messages are stored in the MPFO list 141, the transmission order may change. According to Equation (4) above, F _X [i] depends on F _X [i−1], so if the transmission order is changed, F _X [i] needs to be updated.

Therefore, the control unit 150 recalculates the estimated end times F _X [i] of all messages stored in the queue that have a lower priority than the newly inserted message. Control unit 150 may use the processing time for the transmission of the longest message as P _X [i] when recalculating. In this case, since it is not necessary to distinguish each message stored in the queue 130 waiting for transmission processing, the amount of calculation can be reduced.

In addition, the control unit 150 updates the number Q of active queues as necessary.

-Second Difference The queue 130 waiting to be sent can store pointers to messages to be sent. In that case, the selection of the transmission processing waiting queue 130 in the prioritization processing means that at least one of the MPFO list 141 or the FIFO list 142 is not empty. On the other hand, when a message is sent, the message may be deleted from the transmission queue 130 and the transmission queue 130 may become empty. When the queue 130 waiting for transmission processing becomes empty, the number Q of active queues decreases. In that case, the control unit 150 calculates the virtual time Tvir based on the above equation (3), and overwrites Tvirold and Told with Tvir and Tnow based on the above equation (6).

The differences between the first embodiment and the second embodiment have been described above. An example of routing processing and transmission processing will be described below with reference to FIGS. 8 and 9. FIG.

FIG. 8 is a diagram for explaining an example of message routing processing by the gateway 100 according to the present embodiment. FIG. 8 shows the data structure in gateway 100 and the flow of data following FIG. In the example shown in FIG. 8, if queue #4 is selected by the prioritization process, message 10-3 at the head of queue #4 is stored in the associated MPFO list 141 and the message is deleted from queue #4. be done. Assume that message 10-3 has a lower priority than existing message 10-4 in MPFO list 141. On the other hand, since message 10-3 is stored in MPFO list 141, it is sent before message 10-5 stored in FIFO list 142. FIG. Therefore, the pointer 11-3 to the message 10-3 is inserted between the pointer 11-1 to the top message 10-4 and the pointer 11-2 to the last message 10-5 in the queue #7. be done. Then, the expected end time F ₇ [2] of the message 10-3 is calculated, and the expected end time F ₇ [3] of the message 10-5 is recalculated. Also, since queue #4 is empty, the number of Q is updated.

FIG. 9 is a diagram for explaining an example of message transmission processing by the gateway 100 according to this embodiment. FIG. 9 shows the data structure in the gateway 100 and the data flow following FIG. In the example shown in FIG. 9, queue #7 is selected. A pointer 11-1 to the message 10-4 is stored at the head of the queue #7. Therefore, message 10-4 is stored in transmission buffer 112#6 of channel #6, which is the destination, and is transmitted from channel #6. Also, the transmitted message 10-4 is deleted from the MPFO list 141, and the pointer 11-1 is deleted from the queue #7.

(4) Supplement Prioritization processing, routing processing, and transmission processing can be executed as interrupts or as tasks, but in order to achieve low latency, they are preferably executed as interrupts. This is because interrupts are executed with priority over tasks. However, interrupts for receive processing take precedence over interrupts and tasks for prioritization, routing and transmit processing.

Triggers for prioritization, routing or sending processes can be set using task or interrupt flags. The flag indicates, for example, whether or not there is a non-empty queue or list in the routing queue 120, transmission queue 130, MPFO list 141, and FIFO list 142. In this case, the receive processing interrupt checks the flag and, if the non-empty flag is set, activates the prioritization processing, routing processing, or transmit processing interrupt or task. Furthermore, the interrupts or tasks of the prioritization processing, the routing processing, or the transmission processing are placed in the routing processing waiting queue 120, the transmission processing waiting queue 130, the MPFO list 141, and the FIFO list after executing the prioritization processing, routing processing, or transmission processing. 142 is empty, and if not, call itself again.

The gateway 100 may have a plurality of queues 130 waiting for transmission processing. In that case, the method of selectively using the plurality of transmission processing waiting queues 130 is arbitrary. For example, the plurality of transmission queues 130 may store messages destined for different channels. Also, the plurality of transmission processing waiting queues 130 may alternately store messages in the order in which the routing processing is completed.

<3.3. Process Flow>
(1) Flow of Reception Processing FIG. 10 is a flowchart showing an example of the flow of reception processing executed in the gateway 100 according to this embodiment. As shown in FIG. 10, when a message arrives on channel #X, the controller 150 buffers the incoming message in the receive buffer 111#X corresponding to channel #X (step S302). Next, the control unit 150 calculates the virtual time Tvir by Tvir=Tvirold+(Tnow-Told)/Q (step S304). Next, the control unit 150 moves the message buffered in the reception buffer 111#X to the routing process waiting queue 120 (queue #X) corresponding to the channel #X (step S306).

Next, the control unit 150 determines whether or not the queue #X was empty before the message was moved (step S308). If it is determined to be empty (step S308/YES), the control unit 150 sets A _X [1]=Tvir (step S310), and sets F _X [1]=A _X [1]+P _X [1]. Calculate (step S312) and increment Q (step S314). The process then proceeds to step S320. On the other hand, if it is determined that it is not empty (step S308/NO), the control unit 150 sets A _X [i]=Tvir (step S316), F _X [i]=max(F _X [i−1] , A _X [i])+P _X [i] (step S318). The process then proceeds to step S320.

In step S320, the control unit 150 sets Tvirold=Tvir and Told=Tnow (step S320). Then, the control unit 150 triggers prioritization processing, routing processing, or transmission processing (step S322).

(2) Flow of Prioritization Processing, Routing Processing, and Transmission Processing FIGS. 11A and 11B are flowcharts showing an example of the flow of prioritization processing, routing processing, and transmission processing executed by the gateway 100 according to the present embodiment. is. As shown in FIG. 11A, first, when the prioritization process, the routing process, and the transmission process are triggered by the reception process (step S402), the control unit 150 causes the top of the routing process waiting queue 120 and the transmission process waiting queue 130 to selects the queue with the earliest expected end time F _X [i] of messages (step S404). Let X be the index of the queue to be selected. Next, the control unit 150 determines whether the selected queue #X is the routing process waiting queue 120 or the transmission process waiting queue 130 (step S406). If the selected queue #X is the routing process waiting queue 120, the process proceeds to step S408. On the other hand, if the selected queue #X is the transmission processing waiting queue 130, the process proceeds to step S432. Note that these processes correspond to prioritization processes.

In step S408, control unit 150 selects the first message from queue #X. Next, the control unit 150 determines whether the selected message is an MPFO message or a FIFO message (step S410). If the selected message is a FIFO message, the control unit 150 performs routing and moves the message to the FIFO list 142 (step S412). On the other hand, if the selected message is an MPFO message, the control unit 150 performs routing and moves the message to the MPFO list 141 (step S414). Here, a pointer to the message after being moved to the MPFO list 141 or FIFO list 142 is stored in the queue 130 waiting for transmission processing. Let M be the index of the queue 130 waiting for transmission processing. After step S412 or S414, the process proceeds to step S416. In step S416, control unit 150 calculates virtual time Tvir by Tvir=Tvirold+(Tnow-Told)/Q. Next, the control unit 150 determines whether or not the queue #M was empty before the message was moved (step S308). If it is determined to be empty (step S418/YES), control unit 150 sets A _M [1]=Tvir (step S420), and sets F _M [1]=A _M [1]+P _M [1]. Calculate (step S422) and increment Q (step S424). The process then proceeds to step S430. On the other hand, if it is determined that it is not empty (step S418/NO), the control unit 150 sets A _M [i]=Tvir (step S426), F _M [i]=max(F _M [i−1] , A _M [i])+P _M [i] (step S428). The process then proceeds to step S430. In step S430, the control unit 150 sets Tvirold=Tvir and Told=Tnow. The process then proceeds to step S440. Note that these processes correspond to routing processes.

As shown in FIG. 11B, in step S432, control unit 150 determines whether MPFO list 141 is empty. If it is determined that the MPFO list 141 is not empty, the control unit 150 selects the highest priority message from the MPFO list 141 (step S434). After that, the process proceeds to step S438. On the other hand, if it is determined that the MPFO list is empty, control unit 150 selects the first message from FIFO list 142 (step S436). After that, the process proceeds to step S438. In step S438, the control unit 150 transmits the message to the destination channel determined by the routing process, deletes the transmitted message from the MPFO list 141 or the FIFO list 142, and removes the pointer from the transmission process waiting queue 130. delete. Along with this, the pointer to the deleted transmitted message is deleted from the queue 130 waiting for transmission processing. The process then proceeds to step S440. In step S440, control unit 150 determines whether or not queue #X is empty. If it is determined that queue #X is not empty, the process proceeds to step S446. On the other hand, if queue #X is determined to be empty, control unit 150 calculates virtual time Tvir by Tvir=Tvirold+(Tnow-Told)/Q (step S442). Next, the control unit 150 sets Tvirold=Tvir, Told=Tnow, and Q=Q−1 (step S444). After that, the process proceeds to step S446. Note that these processes correspond to the transmission process.

In step S446, control unit 150 determines whether or not all queues are empty. If it is determined that at least one is not empty, the process returns to step S404. On the other hand, if it is determined that all are empty, the process ends.

<<4. About the comparative example>>
In the following, the gateway 100 described above is compared with a typical gateway.

(1) Description of Comparative Example FIG. 12 is a diagram showing an example of a configuration of a gateway 200 according to a comparative example. As shown in FIG. 12 , the gateway 200 according to the comparative example includes a 6-channel network interface 210 , an MPFO list 241 and a FIFO list 242 . Gateway 200, unlike gateway 100, does not include receive buffers, transmit buffers, and queues. When a message arrives, gateway 200 performs message routing processing and stores the message in a data structure corresponding to the message. Gateway 200 then processes the transmission of the stored message. In the example shown in FIG. 12 , when message 10 arrives on channel # 1 , gateway 200 performs routing processing for message 10 and stores the message in MPFO list 241 . Gateway 200 then transmits the message stored in MPFO list 241 from channel #6.

A simple solution to avoid message loss in gateway 200 is to equip it with a sufficiently powerful computer (hereafter CPU). For example, in a 500 kbps serial bus CAN, a 1-byte message, which is the shortest message, is received in about 100 [μs] (microseconds). That is, it takes 100 [μs] for the gateway 200 to receive from the first bit to the last bit of a 1-byte message. If the gateway 200 is equipped with a CPU capable of processing all received messages within 100 [μs], message loss can be avoided.

On the other hand, an 8-byte message, which is the longest message, is received in approximately 200 [μs]. In order to avoid buffer overrun, it is desirable that even an 8-byte message can be processed within 100 [μs]. This is because a 1-byte message can arrive while an 8-byte message is being processed. Thus, the worst case is one message per channel arriving, each message having the maximum length in CAN (8 bytes), and message processing starting when the first bit of the incoming message is received. is the case. The condition for avoiding message loss in the worst case is, for example, if the gateway 200 has 8 channels, the CPU processes an 8-byte message within 100/8=12.5 [μs].

However, with a typical number of channels (6 to 8 CAN channels) and a typical automotive CPU (with a clock speed of about 100 MHz), it is difficult to satisfy the above conditions. In order to satisfy the above conditions, it is desirable to adopt a powerful CPU with a clock speed of about 200 MHz and/or with multiple cores. However, employing such a powerful CPU has a negative impact on cost. In CAN-FD, this adverse effect is exacerbated by the high data rates and long messages.

Moreover, even if a sufficiently powerful CPU were employed to avoid message loss, there would still be the possibility of delays in sending messages. This is because, under heavy bus loads, the gateway 200 may prioritize message reception over transmission and store received messages without transmitting them. For example, in an over-the-air reprogramming session, gateway 200 may receive a large number of low priority reprogramming messages while also receiving high priority control messages. Immediate transmission of high priority control messages becomes difficult when message reception is prioritized to avoid message loss. Such a situation is absolutely unacceptable in a real-time system.

It is also conceivable to prioritize sending messages over receiving messages. However, in that case it may be difficult to avoid message loss. For example, in the wireless reprogramming session example above, gateway 200 may have a large buffer of low priority reprogramming messages waiting to be sent, while failing to receive high priority control messages.

A fine balance between receiving and sending messages at the gateway 200 is required to meet the demands of real-time systems such as in-vehicle systems. And it is difficult to achieve this delicate balance with a simple strategy such as giving priority to receiving or sending messages.

(2) Comparison with Comparative Example A gateway 100 according to an embodiment (first or second embodiment) of the present invention has M queues, which are larger than the number of channels N, and each of the M queues Allocate computational resources to messages selected according to a predetermined prioritization algorithm from the messages stored in . This relaxes the requirements to avoid message loss. A condition for avoiding message loss is to process messages received in x[μs] within x/M[μs].

For example, if the gateway 100 has 8 channels and 9 queues and receives a 1-byte message within 100 [μs], it is required to process the message within 100/9=11.1 [μs]. be done. This processing time is at the same level as gateway 200 . On the other hand, if the gateway 100 has 8 channels and 9 queues and receives an 8-byte message within 200 [μs], it is required to process the message within 200/9=22 [μs]. . This processing time is approximately double that of gateway 200 . In other words, the gateway 100 can avoid message loss even with a CPU having about half the processing power of the CPU installed in the gateway 200 .

Further, in this embodiment, (MN)/M of computational resources are allocated for transmission of messages. For example, in an 8-channel gateway with 9 queues, 11% of computational resources (eg, CPU processing time) are spent sending messages. Of course, the number of queues can be adjusted arbitrarily depending on the number of channels and the desired percentage of the CPU's computational resources to allocate to sending messages.

In this way, the gateway 100 can avoid message loss even with low processing power (ie, low cost) CPUs. In addition, the gateway 100 can secure transmission opportunities even when the load of received messages is high.

<<5. Supplement >>
Although the preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings, the present invention is not limited to such examples. It is obvious that a person having ordinary knowledge in the technical field to which the present invention belongs can conceive of various modifications or modifications within the scope of the technical idea described in the claims. It is understood that these also naturally belong to the technical scope of the present invention.

Note that the gateway 100 described herein may be implemented as a single device, or partially or wholly may be implemented as separate devices. For example, in the functional configuration example of the gateway 100 shown in FIG. In this case, information indicating the states of the network interface 110, the routing process waiting queue 120, the transmission process waiting queue 130, and the third storage unit 140 and the stored messages is sent to the control device. Then, the control unit 150 in the control device performs various processes on the received information.

Also, the series of processes by the gateway 100 described herein may be implemented using software, hardware, or a combination of software and hardware. Programs constituting software are stored in advance in a storage medium (non-transitory media) provided inside or outside the gateway 100, for example. Then, for example, the program is read into a RAM when executed by a computer, and executed by a processor such as a CPU.

Also, the processes described in this specification using flowcharts and sequence diagrams do not necessarily have to be executed in the illustrated order. Some processing steps may be performed in parallel. Also, additional processing steps may be employed, and some processing steps may be omitted.

The references cited above are as follows.
[1] Sergio Campoy-Martinez, Vincent Bidault: “Method and device for transmitting frames between two CAN buses”, WO2014127866 (A1), Renault SAS, priority date 2013/02/21
[2] John Nagle: "On packet switches with infinite storage," RFC 970, IETF, December 1985.
[3] Albert Greenberg, Neal Madras: "How fair is fair queuing?", Journal of the ACM, Volume 39 Issue 3, July 1992, pp. 568-598.

REFERENCE SIGNS LIST 100 gateway 110 network interface 111 fourth storage unit, reception buffer 112 fifth storage unit, transmission buffer 120 first storage unit, routing processing waiting queue 130 second storage unit, transmission processing waiting queue 140 third storage Part 141 MPFO list 142 FIFO list 150 Control part 200 Gateway according to the comparative example

Claims (11)

An in-vehicle gateway device (100) connected to N networks and relaying messages between different networks,
N first storage units (120) for storing messages awaiting routing processing among messages received from each of the N networks;
one or more second storage units (130) that store information indicating messages that have been routed and are waiting for transmission;
a control unit (150) that selects one message as a processing target from the messages waiting for routing processing and the messages waiting for transmission processing, and allocates computational resources to the routing processing or the transmission processing of the message to be processed; ,
with
The control unit (150) selects one storage unit (120 or 130) from the first storage unit (120) and the second storage unit (130), ) is selected as the processing target, and
The control unit (150) selects one storage unit (120 or 130) from the non-empty first storage unit (120) and the second storage unit (130) in a round-robin fashion.
An in-vehicle gateway device (100). 2. The control unit according to claim 1 , wherein said control unit calculates an estimated end time of processing of each of said message waiting for routing processing and said message waiting for transmission processing, and selects a message with the earliest estimated end time as said processing target. An in-vehicle gateway device (100) as described. The control unit controls the scheduled end time of a message to be processed before the message to be calculated, which is stored in the same storage unit (120 or 130) as the message to be calculated, or the message to be calculated. calculating the scheduled end time of the message to be calculated by adding the processing time of the message to be calculated to the later one of the times when the message was stored in the storage unit (120 or 130); Item 3. The in-vehicle gateway device (100) according to item 2. The control unit controls the later of the scheduled end time or the virtual time of a message that is stored in the same storage unit (120 or 130) as the message to be calculated and that is processed before the message to be calculated. calculating the scheduled end time of the calculation target message by adding the processing time of the calculation target message to the calculation target message,
The virtual time stores the difference between the current real time and the real time at the last update of the virtual time in one of the first storage unit (120) and the second storage unit (130) that is not empty. 3. The in-vehicle gateway device (100) according to claim 2 , updated by adding a value obtained by dividing by the number (120 or 130) to said virtual time at the previous update. 5. The in-vehicle gateway device (100) according to claim 3 or 4 , wherein the processing time depends on the size of the message to be calculated. The in-vehicle gateway device (100) according to any one of claims 3 to 5 , wherein said processing time depends on the type of processing applied to said message to be calculated. The in-vehicle gateway device (100) further comprises a third storage unit (140) for storing the messages waiting for transmission processing,
The second storage unit (130) according to any one of claims 1 to 6 , wherein the second storage unit (130) stores pointers to the messages waiting for transmission processing stored in the third storage unit (140). An in-vehicle gateway device (100). 8. The in-vehicle gateway device (100) according to claim 7 , wherein said third storage unit (140) includes at least one of an MPFO list (141) and a FIFO list (142). further comprising N fourth storage units (111) for storing messages received from each of the N networks;
The control unit allocates the computing resources to the processing of moving the message stored in the fourth storage unit (111) to the first storage unit (120), prior to the routing processing and the transmission processing. In-vehicle gateway device (100) according to any one of claims 1 to 8 , allocating. A method performed by an in-vehicle gateway device (100) connected to N networks and relaying messages between different networks, comprising:
a step (S104 or S306) of storing messages awaiting routing processing among messages received from each of the N networks in N first storage units (120);
a step of storing in one or more second storage units (130) information indicating messages that have been routed and are awaiting transmission processing (S212, S214, S412, or S414);
Selecting one message as a processing target from the messages waiting for routing processing and the messages waiting for transmission processing, and allocating computational resources to the routing processing or the transmission processing of the message to be processed (S208, S218, S222, S408, S434 or S436 and S212, S214, S224, S412, S414 or S438);
including
In the step of allocating computational resources to the routing process or the transmission process of the message to be processed,
selecting one storage unit (120 or 130) from the first storage unit (120) and the second storage unit (130), and processing the message corresponding to the selected storage unit (120 or 130) Select as target and
selecting one said storage unit (120 or 130) in a round-robin fashion from said first storage unit (120) and said second storage unit (130) that are not empty;
Method. A computer that controls an in-vehicle gateway device (100) that is connected to N networks and relays messages between different networks,
a step (S104 or S306) of storing messages awaiting routing processing among messages received from each of the N networks in N first storage units (120);
a step of storing in one or more second storage units (130) information indicating messages that have been routed and are awaiting transmission processing (S212, S214, S412, or S414);
Selecting one message as a processing target from the messages waiting for routing processing and the messages waiting for transmission processing, and allocating computational resources to the routing processing or the transmission processing of the message to be processed (S208, S218, S222, S408, S434 or S436 and S212, S214, S224, S412, S414 or S438);
and
In the step of allocating computational resources to the routing process or the transmission process of the message to be processed,
One storage unit (120 or 130) is selected from the first storage unit (120) and the second storage unit (130), and the message corresponding to the selected storage unit (120 or 130) is processed. selected as a target
One said storage unit (120 or 130) is selected in a round-robin fashion from said non-empty first storage unit (120) and said second storage unit (130),
program.

Images