Method and corresponding apparatus for the automatic detection of the bit rate of a network, particularly of type CAN (Controller Area Network), and for the configuration of the network at the detected bit rate

The field of the invention is that of communications networks.

More specifically, the invention relates to an automatic detection method, a device connected to a network, the flow rate of the network and device configuration the sensed flow. The invention also relates to the device implementing this process.

The invention is particularly, but not exclusively, in case the CAN bus network is kind.

By way of illustration and for the sake of simplicity, we now present the technique of the prior art and its drawbacks in the aforementioned case where the network is CAN-bus.

Recall that a CAN ( "Controller Area Network") is a serial bus for connecting a plurality of devices, called "control units" each including a CAN controller. This type of bus is now mainly used in the field of industry and in the automotive field.

Typically, in a car, two CAN buses are used. One said CAN bus High speed (up to 1 Mb / s), used to interconnect related control units or to the following features: dashboard, engine, braking (anti-lock) active suspension, transmission, etc. The other, said low-speed CAN bus (Up to 125 kb / s), used to interconnect elements related control units or following features: dashboard, lighting, air conditioning, cushions Safety airbags, door locks, power windows, etc.

Each CAN bus implements a protocol of the same name (CAN protocol) which is a serial communication protocol that supports real-time systems a high level of reliability in a limited and harsh environment like a factory, workshop, a car ... The CAN protocol covers two of the seven layers of the model of OSI OSI ISO systems, namely the physical layer (Layer 1) and data link layer (layer 2). For more information on the CAN bus, one may refer to ISO 11898, inserted by reference.

The CAN protocol data link layer is such that each unit Control can transmit and receive data. Data is transmitted on the bus in the form of packets (also called frames or messages) asynchronous format defined but variable and limited length. As soon as the bus is free, any control unit connected to the bus may transmit a new packet. a mechanism interruption for the priority packets is provided, and a mechanism arbitration of disputes arising from the simultaneous transmission of multiple packets on the bus when it is free.

Traditionally, when you want to connect to a CAN bus new control unit (that is to say a new device including a CAN controller running an application (layer 7 of the OSI model)), the procedure is as follows. The developer of the application executed by the CAN controller needs to know the frequency CAN controller clock (eg 12 MHz). It also assumes that the flow network takes one of a limited number of values ​​(three values ​​generally). By example, it assumes that the network throughput is 100, 250 or 500 kb / s. For each three of assumed values ​​of the network throughput, it determines a priori, and from the Frequency knowledge CAN controller clock, a flow configuration for the CAN controller. It then develops the application so that, in operation, application load a first rate configuration associated with a first assumed value of the network throughput; If the application receives messages error, it loads a second rate configuration associated with a second assumed value of the network throughput; and so on in order to test (if necessary) predetermined different speed configurations.

A drawback of the aforesaid known art is that it can not be used in situations where we do not know the frequency of the controller clock CAN.

Another drawback of the aforesaid known art is that it does not work if no assumptions about the network speed is correct.

Yet another drawback of the abovementioned known technique is that when several possible flow configurations are tested successively, the controller CAN really disturb the CAN bus. This can even lead to ejection of CAN controller CAN bus.

The invention particularly aims to overcome these various disadvantages of the state of the art.

More specifically, one of the objectives of the present invention is to provide a method and device for automatically detecting the throughput of a network and configuration at the sensed flow rate, does not require knowledge of the frequency device clock or assumptions about the network throughput.

The invention also aims to provide such a method and device not disrupting the network.

Another object of the invention is to provide such a method and device which are simple to implement and inexpensive.

These various objectives, and others that will appear later, are achieved according to the invention by means of an automatic method for detecting the flow rate of a network configuration and the speed detected by a device connected to the network, method comprising:<ul><li>an initialisation phase comprising the steps of:<ul><li>the device is placed in a listening mode, where it can receive messages from the network but can not send messages on network;</li><li>the device is initialized with an initial rate configuration;</li><li>a first timer is started;</li></ul></li><li>an iterative processing step comprising, for each iteration, the steps following:<ul><li>if at least one rate matching condition is satisfied, is interrupted phase iterative processing, it validates the configuration of current flow, and Place the device in a normal mode;</li><li>if at least one flow mismatching condition is satisfied or if the first timer expires without a rate matching condition or a failure to adapt to flow condition has been verified, we reset the First timer and current flow configuration is replaced by a new flow pattern obtained by implementing a mechanism obtaining a plurality of possible flow configurations without Knowledge of a clock frequency f_h of the device and without hypothesis on the network throughput.</li></ul></li></ul>

The general principle of the invention therefore is to test, in a listening mode to not disturb the network, a plurality of flow configurations obtained without knowledge of the clock frequency of the device and no assumption the network throughput.

Advantageously, said at least one non-rate matching condition is the detection by the device of an error, said error bit on at least one bit of a message received by the device.

Advantageously, said at least one throughput adaptation condition is acknowledgment by the device of a received message.

Preferably, said mechanism provides successively all possible flow configurations of the device.

According to an advantageous characteristic, the initialization phase comprises further a step of initiating a second time delay, and the phase iterative processing further comprises, for each iteration, a step of interrupting the process if the second timer expires without a configuration of current flow has been validated.

In one advantageous embodiment of the invention, the network is a bus CAN, and the device includes a CAN controller completed compared to controller CAN vector so as to implement the method according to the invention detection of Automatic Network flow and the sensed flow configuration.

Each of the bits included in a message for example has a time nominal value defined by the following parameters:<ul><li>scalar BRP, such as: TQ = BRP / f_hWith TQ a unit of time and f_h the device clock frequency;</li><li>the length L_PRS, Expressed in number of TQ unit time, of a segment spread to compensate for the physical delay on the CAN bus; </li><li>the lengths L_PHS1 and I_PHS2Expressed in number of TQ time unit, of first and second phase segments allowing to position a point sampling.</li></ul>

In this case, advantageously, for each of the plurality of patterns of possible flow, said mechanism determines values ​​for the scalar BRP and lengths L_PRS, L_PHS1 and I_PHS2Without knowledge of the clock frequency f_h of the device and without hypothesis on network throughput.

Advantageously, said mechanism implements an incrementing loop of the length L_PHS2, Itself included in an incrementing loop length L_PHS1, Itself included in an incrementing loop length L_PRS, herself included in a scalar increment loop BRP. Buckles incrementation of lengths L_PRS, L_PHS1 and I_PHS2 are such that the lengths L_PRS, L_PHS1 and I_PHS2 are incremented evenly.

The invention also relates to a device intended to be connected to a network, characterized in that it comprises means for automatic detection of the flow network configuration and the detected rate, themselves including:<ul><li>initialization means comprising:<ul><li>tilting means for a normal mode to a listening mode, in which are activated for locking means of messages transmitted by the device to the network, the device continuing to receive messages from the network;</li><li>initialization means of the device with an initial rate configuration;</li><li>triggering means for a first time delay;</li></ul></li><li>iterative processing means comprising the following means, activated when each iteration:<ul><li>first detecting means that at least one matching condition of Speed ​​is checked, the following means being activated in case of positive detection by the first detecting means: means for interruption of the treatment iterative, enabling means of the configuration of current flow and tilting means listening mode to normal mode;</li><li>second detecting means at least one non-matching condition flow is verified or that the first timer expires without a rate adaptation condition or debit mismatching condition has been verified, the following means being activated in case of positive detection by second detecting means: means for resetting the first and delay means for replacement of the rate configuration current by a new configuration of flow obtained by implementation of a mechanism for obtaining a plurality of possible flow configurations without Knowledge of a clock frequency f_h of the device and without hypothesis on the network throughput.</li></ul></li></ul>

Advantageously, said at least one non-rate matching condition is the detection by the device of an error, said error bit on at least one bit of a message received by the device.

Advantageously, said at least one throughput adaptation condition is acknowledgment by the device of a received message.

Preferably, said mechanism provides successively all possible flow configurations of the device.

According to an advantageous characteristic, the initialization means include further means for triggering a second timer. In addition, iterative processing means further comprises third detecting means, activated during each iteration, for detecting whether the second timer expires without a configuration of current flow has been validated, means interrupt the process being activated in the event of positive detection by the third detection means.

In one advantageous embodiment of the invention, the network is a bus CAN, and the device includes a CAN controller completed including himself automatic said detection means of the network throughput and the speed configuration detected.

Each of the bits included in a message for example has a time nominal value defined by the following parameters:<ul><li>scalar BRP, such as: TQ = BRP / f_hWith TQ a unit of time and f_h the device clock frequency; </li><li>the length L_PRS, Expressed in number of TQ unit time, of a segment spread to compensate for the physical delay on the CAN bus;</li><li>the lengths L_PHS1 and I_PHS2Expressed in number of TQ time unit, of first and second phase segments allowing to position a point sampling.</li></ul>

In this case, advantageously, for each of the plurality of patterns of possible flow, said mechanism determines values ​​for the scalar BRP and lengths L_PRS, L_PHS1 and I_PHS2Without knowledge of the clock frequency f_h of the device and without hypothesis on network throughput.

Advantageously, said mechanism implements an incrementing loop of the length L_PHS2, Itself included in an incrementing loop length L_PHS1, Itself included in an incrementing loop length L_PRS, herself included in a scalar increment loop BRP. Buckles incrementation of lengths L_PRS, L_PHS1 and I_PHS2 are such that the lengths L_PRS, L_PHS1 and I_PHS2 are incremented evenly.

Advantageously, the tilting means of the normal mode listening means comprise generation of a mode selection signal (Autobaud), taking the value "0" or "1" depending on whether the device is to operate in normal mode or listening respectively. In addition, the CAN controller completed includes a conventional CAN controller modified to include himself said automatic means for detecting the network throughput and configuration to the detected flow, except said transmitted message blocking means. Finally, the locking means transmitted messages comprise means for realization of the logic function "OR" between the mode selection signal and the transmission signal (TxDC ') of the CAN controller classic, the output signal of the means for implementing the logic function "OR" constituting the transmission signal (TXDC) completed the CAN controller.

Advantageously, the completed CAN controller further comprises means loopback output to the input of conventional CAN controller, including:<ul><li>ways to achieve the logic function "AND" between the signal Issue (TxDC ') of conventional CAN controller and the reception signal (RXDC) completed the CAN controller; </li><li>multiplexing means receiving at a first input the output signal ways to achieve the logic function "AND" and a second input signal reception (RXDC) completed the CAN controller, signal output multiplexing means constituting the receive signal (RXDC ') of conventional CAN controller, the multiplexing means being controlled by the mode selection signal such that the first or the second input is selected by the mode select signal is set to "1" or "0" respectively.</li></ul>

Other features and advantages of the invention will become apparent from reading the following description of a preferred embodiment of the invention, given as an indicative and non-limiting example and the accompanying drawings, wherein:<ul><li>1 shows a block diagram of a known system of interconnection of plurality of devices via a CAN bus system wherein can be implemented the method according to the invention detection and configuration Automatic flow;</li><li>2 illustrates the known structure of each bit included in a message carried on a CAN bus;</li><li>the Figure 3 shows a functional block diagram of an exemplary embodiment of a device shown in Figure 1 and for carrying out the method according to the invention;</li><li>4 shows a particular embodiment of the adapter block electronic appearing in FIG 3;</li><li>5 shows a flowchart of a particular embodiment of the method of the invention.</li></ul>

The invention therefore relates to a method and a device for automatically detecting the flow of a network configuration and the sensed flow.

In the following description, one considers only the case where the network is a CAN bus. It is clear however that the invention is also applicable to other types network.

As already mentioned above and illustrated in the block diagram of <b><u>figure 1</u></b>, a bus 1 CAN is a serial bus for connecting a plurality of devices 2₁, 2₂... 2_not, called "control units" and each including a CAN controller.

Now we briefly recall some essential characteristics of CAN protocol, which is a serial communication protocol implemented by a bus CAN. For more information on the CAN protocol and the CAN bus, reference may be Refer to ISO 11898, inserted by reference.

The CAN protocol distinguishes four types of CAN messages:<ul><li>the data message (Data Frame), the standard format (11-bit identifier) or extended (29-bit identifier);</li><li>the data request message (Remote Frame);</li><li>error messages (Error Frame);</li><li>network delay message (Overload Frame).</li></ul>

A standard CAN message includes:<ul><li>a start of frame field;</li><li>an arbitration field, which consists of an ID and a bit "Data Request" used to distinguish a data message (Data Frame), a request message (Remote Frame);</li><li>a control field, which contains an "extended identifier" bit (IDE) and a number of data (DLC) indicating the number of data byte which in the following data field. In a request message, the number DLC is the number of requested data byte;</li><li>a data field of up to eight bytes of data;</li><li>a check (CRC) field, ensuring message integrity;</li><li>an acknowledgment field (ACK), consisting of a slot and a delimiter acknowledgment. The bit in the acknowledgment segment is transmitted at a level recessive and crashed at a dominant level by receiving devices had at that time received the data correctly;</li><li>a field message end (EOF);</li><li>an inter-message field (IFS), which is a minimum number of bits between two consecutive messages.</li></ul>

The CAN protocol distinguishes five types of errors, which are not mutually exclusive mutually:<ul><li>a bit error ( "bit error") when the bit itself is tainted Error;</li><li>a jam error ( "bit stuff error");</li><li>an error acknowledgment ( "ACK error");</li><li>a verification error ( "CRC error");</li><li>a structural error.</li></ul>

A device connected to the CAN bus and detects an error condition has Mission report by sending an error message. These features are a device that plugs into a CAN bus must have a flow configuration OK, at the risk of the existing CAN bus or at risk of being excluded from this bus CAN.

Will now be described in connection with the <b><u>2</u></b>, The known structure ( "CAN Bit timing "), and defined in the CAN protocol, each bit included in a message conveyed on a CAN bus.

The nominal time for each bit (or NTB, for "Nominal Bit Time" in English) is defined by a structure composed of four distinct segments 21-24. Each of these four segments is composed of an integral number of unit time (or TQ for "Time Quanta" in English). The length of this unit of time TQ is defined from the clock frequency f_h of the device and a scalar BRP to own device: TQ = BRP / f_h. The total number of time units TQ in one bit must be between 8 and 25.

The first segment is a sync segment 21 ( "synchronization segment ") whose length is fixed to a unit of time TQ.

The second segment 22 is a propagation segment ( "segment spread"). Its role is to compensate for the physical delay induced by the CAN bus. Its length L_PRSis programmable between one and eight time units TQ.

The third and fourth segments 23, 24 ( "Segment Phase 1 'and' period Segment 2 ') are phase segments, used to position a point sampling. Their lengths L_PHS1 and I_PHS2 are also programmable.

It has now, in connection with the <b><u>3</u></b>, An embodiment a device shown in Figure 1 (for example the one referenced 2₁) And to the implementation of the method according to the invention.

Usually, the device 2₁ includes a controller (or microcontroller) CAN classic 31 and a transmitter / receiver circuit 33 CAN ( "CAN transceiver"). This last 33 is specific to the physical layer CAN protocol and is designed to the interface between the conventional CAN controller 31 (which works to logic) and the bus CAN 1 (who works at the physical level). The conventional CAN controller 31 outputs a signal 35 (hereinafter referred to as transmission signal TxDC ') and receives another 36 (Hereinafter referred to as reception signal RXDC ').

According to the present invention, the conventional CAN controller 31 is amended in than :<ul><li>it executes a detection algorithm and automatic configuration flow (process of the invention, described in detail below in connection with Figure 5);</li><li>it generates a mode selection signal 37 (Autobaud).</li></ul>

Device 2₁ is also modified in that it further comprises a block Electronic adaptation 32, interfacing between the conventional CAN controller 31 and transmitter / receiver circuit 33. The conventional CAN CAN controller 31 and block Electronic adjustment 32 together form a CAN controller completed 34 own to the present invention. Thus, the electronic adaptation block 32:<ul><li>on one side, receives the transmission signal TxDC "and the selection signal fashion 37 (Autobaud) emitted by the conventional CAN controller 3, and outputs the receiving signal RXDC 'received by the conventional CAN controller 3;</li><li>each other, transmits the transmission signal TxDC CAN controller completed 4 and receives the reception signal RXDC 'CAN controller completed 4.</li></ul>

The electronic adjustment unit 32 is controlled by the selection signal Mode 37 (Autobaud) so that, during the execution of the above algorithm, the 2 device₁ switches from one to the other of the following modes:<ul><li>a normal mode in which the conventional CAN controller can issue 31 or receive messages to or from the CAN bus; </li><li>a listening mode, in which the conventional CAN controller 31 may receive messages from the CAN bus but can not transmit on This CAN bus, and in which the conventional CAN controller 31 can receive any messages that it sends to the CAN bus (and as indicated previously are blocked before they get there).</li></ul>

Will now be described in connection with the <b><u>4</u></b>, An embodiment particular electronic matching block 32 appearing in FIG 3.

It includes messages locking means emitted by the CAN controller 31 classic and loopback means of the output to the input of the CAN controller 31 classic.

The mode selection signal 37 (Autobaud) takes the value "0" if the 2 device₁ must operate in normal mode, or "1" for operation in the listening mode.

The transmitted message blocking means comprises an OR gate 41, realizing the logic function "OR" between the mode selection signal 37 (Autobaud) and the transmission signal (TxDC ') of the conventional CAN controller 31. The output signal OR gate 41 is the transmission signal (TxDC) CAN controller completed 34.

Thus, when the mode selection signal 37 (Autobaud) takes the value "0" (Normal mode), the transmission signal (TxDC ') of conventional CAN controller 31 is confused with the transmission signal (TxDC) completed the CAN controller 34, and 2 device₁ can therefore transmit messages on the CAN bus.

However, when the mode selection signal 37 (Autobaud) is set "1" (listen mode), the transmission signal (TxDC) CAN controller completed 34 is always "1", which corresponds to a device of the resting state (as "1" the recessive state of the CAN bus) and therefore no message is sent by the device 2₁ on the CAN bus.

the output of the loop means to the input of conventional CAN controller 31 include:<ul><li>an AND gate 42, producing the logical "AND" function between the transmission signal (TxDC ') of conventional CAN controller 31 and the reception signal (RXDC) of CAN controller completed 34; </li><li>a multiplexer 43 receiving at a first input the output signal of the AND gate 42 and on a second input the received signal (RXDC) of CAN controller completed 34. The multiplexer output signal is the reception signal (RXDC ') of conventional CAN controller 31. The multiplexer is controlled by the mode selection signal 37 (Autobaud), so that the first or the second input is selected depending on whether the selection signal mode is set to "1" or "0" respectively.</li></ul>

Thus, when the mode selection signal 37 (Autobaud) takes the value "0" (Normal mode), the received signal (RXDC ') of conventional CAN controller 31 is confused with the reception signal (RXDC) completed the CAN controller 34, and 2 device₁ therefore receives messages from the CAN bus.

However, when the mode selection signal 37 (Autobaud) is set "1" (listen mode), the received signal (RXDC ') of conventional CAN controller 31 is a combination (AND function) of the transmission signal (TxDC ') of the CAN controller Classic 31 and the reception signal (RXDC) CAN controller completed 34. Due "1" is the recessive state of the CAN bus, this allows to enjoy both messages from the CAN bus as messages (error message or messages including acknowledgment) emitted by the conventional CAN controller 31.

Now be presented in connection with the flowchart <b><u>5</u></b>, a particular embodiment of the process according to the invention and detection of Automatic flow configuration.

One distinguishes an initialization phase 51 and an iterative processing step 52.

Each flow configuration is defined by the values ​​given to the parameters follows (see the above description of Figure 2):<ul><li>the scalar BRP (rated "prescaler" in Figure 5);</li><li>the length L_PRS (Denoted "L propagation segment" in Figure 5) expressed in units of time TQ, the propagation segment 22;</li><li>the lengths L_PHS1 and I_PHS2 (Denoted "L Phase segment 1" and "L segment Phase 2 "respectively in Figure 5), also expressed in units of time TQ, the first and second phase segments 23, 24.</li></ul>

The initialization phase 51 comprises: <ul><li>place the device in the listening mode ( "listening mode" in English) giving the value "1" in the mode selection signal (Autobaud) (see discussion above);</li><li>initialize the device with an initial flow configuration, corresponding to a minimum flow rate value. For this, given by example the initial value "1" to the scalar BRP ( "prescaler"), "3" on length L_PRS ( "The propagation segment") and "2" to each lengths L_PHS1 and I_PHS2 ( "L segment Phase 1" and "L PHASE2 segment");</li><li>trigger (that is to say, delete) a first timer (denoted "Timeout CAN" in Figure 5);</li><li>trigger a second time (denoted "Autobaud timeout" on Figure 5).</li></ul>

The first and second time delays are for example managed by counters that are specific to the CAN controller 34 completed.

La première temporisation (« timeout CAN ») est utilisée pour changer de rate configuration if a first predetermined time after the CAN controller 31 classic has issued (and therefore received since its input is looped to its output) or a Bit error message or an acknowledgment message a received message.

The second delay ( "timeout Autobaud") is used to stop performing the method on a failure if after a second predetermined time period no flow configuration could not be validated. This avoids an endless loop.

now details the steps performed at each iteration of the iterative processing stage 52.

During the step referenced 53, it is detected whether the second timer ( "Timeout Autobaud") has elapsed. If this is the case, it stops execution of the process on a failure ( "autobaud KO"), referenced in step 54.

Otherwise, proceed to the step referenced 55, in which it is detected whether the conventional CAN controller issued a 31 bit error message. If it's not the case, detecting whether the first time delay ( "timeout CAN") has elapsed, in step referenced 56.

If the first timer ( "CAN timeout") has not elapsed, it passes step referenced 57, in which it is detected whether the conventional CAN controller 31 issued a of a previously received message acknowledgment message. If this is the case, when the step referenced 58 is stopped performing the method on a success ( "autobaud OK "), in validating the configuration of current flow and by placing the device in the normal mode. If the conventional CAN controller 31 has not delivered message acknowledgment, one returns to the step referenced 53 (new iteration).

If the conventional CAN controller issued a 31 bit error message or if the first time ( "CAN timeout") has passed, it goes to step referenced 59, in which it resets the first timer ( "timeout CAN"). Then, during the step referenced 510, it is detected whether the scalar BRP ( "prescaler") is less than a predetermined maximum scalar ( "prescaler max"). If this is not the case, proceed to the step referenced 511, in which it is detected whether the current length of the second phase segment ( "segment 2 phase L") is less than the current length of propagation segment ( "The propagation segment"). If the answer to the step referenced 511, we increment the current length of the second phase segment, during the step referenced 512 and returns to the step referenced 53 (new iteration). If the answer to the step referenced 511, proceed to the step referenced 513, in which it is detected whether the current length of the first phase segment ( "The Phase Segment 1 ') is less than the current length of the propagation segment ( "The propagation segment"). If the answer to the step referenced 513, is increments the current length of the first phase segment, at the step referenced 514 and returns to step referenced 53 (new iteration). In case of positive response referenced in step 513, proceed to the step referenced 515, in which we detects whether the current length of the propagation segment ( "segment L spread") is less than eight (the maximum value of this length). If the answer to the step referenced 515, proceed to the step referenced 516, in which we increments the current length of the propagation segment and then returns to step referenced 53 (new iteration). If the answer to the step referenced 515, passes to step referenced 517, in which the value is incremented scalar BRP ( "prescaler") and restores their initial values ​​( "2", "2" and "3" respectively) to the lengths of first and second phase segments ( "L segment Phase 1 "and" L Phase Segment 2 ') and the length of the propagation segment ( "L segment spread ") and then returns to step referenced 53 (new iteration).

If, at the step referenced 510, it is detected that the scalar BRP ( "prescaler") is not less than the maximum scalar ( "prescaler max"), proceed to step referenced 518, in which it restores their initial values ​​( "1", "2", "2" and "3" respectively) to the set of parameters (scalar BRP ( "prescaler") lengths first and second phase segments ( "L segment Phase 1" and "L-phase segment 2 ") and length of the propagation segment (" segment L spread ")), then returns to step referenced 53 (new iteration).

Note that incrementing loop segment length propagation and the lengths of the first and second phase segments are such that these three lengths are increased evenly.

The method according to the invention, detection and automatic flow configuration, can be realized in software (software) or hardware (hardware).