KR100371098B1 - Methode for allocating bandwidth in control area network protocol - Google Patents

Abstract

A method of allocating bandwidth within a Controller Area Network (CAN) protocol is proposed. Real-time event data, real-time control data, and non-real-time data generated from field devices in automation and process control systems commonly share CAN bus media. The method of the present invention divides the bandwidth of the CAN bus into time slots of a macro cycle equal to the minimum time of sampling times between control loops, event data as real time data, control data as real time data and non real time data. Marking an appropriate value in the identifier fields of their data so as to have a transmission priority in that order, subdividing the macro cycle into real-time and non-real-time segments, and retrieving the event data and control data during the real-time segment. And sequentially transmitting the non-real-time data during the non-real-time period. The bandwidth allocation algorithm according to the present invention not only satisfies the performance requirements of real-time application systems interconnected with the CAN, but also makes full use of the bandwidth of the CAN.

Description

How to allocate bandwidth within the CAN protocol {METHODE FOR ALLOCATING BANDWIDTH IN CONTROL AREA NETWORK PROTOCOL}

The present invention relates to a method of allocating bandwidth in a control area network (CAN).

The CAN is a serial communication protocol that supports distributed real-time control and automation. The CAN is used in many industrial process control and automation systems because of its low cost and simplicity. Data generated from distributed control and automation systems is classified as real time and non real time data. On the CAN bus these data share a common network medium. In this situation, the transmission of real time data may be interrupted by non real time data, and the network delay of the real time data may exceed a predefined limit. This causes performance degradation and even instability of real-time application systems interconnected to the CAN bus. Thus, when the CAN bus is instrumented on industrial automation and control systems, its network traffic must be fully analyzed and controlled.

Hereinafter, the CAN protocol will be described.

The CAN protocol operates on the basis of carrier sense multiple access (CAMA / NBA) mechanism. Each node senses the state of the bus before sending data. Nodes postpone the transmission of data until the state of the bus is idle. When an idle condition of the bus is detected, any node can start transmitting. Once more than one node starts sending its message, bus access conflicts should be resolved. This can be done by comparison of identifier fields. The CAN uses one wired OR bus to connect all nodes. While transmitting the identifier, each node monitors the serial bus line. If the transmit bit is weak (logic 1) and the dominant bit (logic 0) is monitored, the nodes give up transmitting and start receiving data. The node sending the highest priority identifier passes through to reach the bus. The identifier for each message must be unique to prevent it from having an equal priority.

An object of the present invention is to provide a bandwidth allocation algorithm that can be applied to the CAN protocol.

Another object of the present invention is to provide a method for allocating bandwidth in a CAN protocol that can efficiently utilize traffic of the CAN protocol to ensure the performance of real-time application systems connected to the CAN bus and at the same time make full use of it. have.

In order to achieve these objects, according to the invention, the window scheduling algorithm is extended to the bandwidth allocation of the CAN protocol to accommodate real time and non real time data.

1 is a diagram illustrating the allocation of bandwidth within a CAN bus.

2 is a table showing the shapes of nodes in a simulation model.

3 is a table comparing the performance of algorithms and algorithms at B = 125KB / s.

4 is a table comparing the performance of algorithms and non-algorithms at B = 100 KB / s.

5 (A) and (B) are graphs showing the performance of the control system.

6 is a graph showing the number of waiting messages in a transmission queue to verify stabilization conditions of non-real-time data.

A method of one embodiment of the present invention has a control area network bus (CAN) and a plurality of nodes and a plurality of control loops sharing this CAN bus, each control loop having at least one data transmission node, a sensor and a controller. A method of allocating a bandpole in a CAN protocol comprising: dividing the bandwidth of the CAN bus into time slots of a macro cycle equal to a minimum time of sampling times between the control loops, wherein the bandwidth is real time data. Marking an appropriate value in the identifier fields of their data in order of event data, control data that is real-time data, and non-real-time data, subdividing the macro cycle into real-time and non-real-time segments, and The event data and the control data in this order during the real time period Transmitted to, and a step of transmitting the non-real-time data in said non-real time interval.

On the other hand, the method according to another aspect of the present invention has a CAN bus and a plurality of nodes and a plurality of control loops sharing the CAN bus, each control loop comprising at least one data transmission node, a sensor and a controller. A method for allocating bandwidth within a CAN protocol comprising: M: number of control loops interconnected to a CAN bus, Nc: number of nodes generating control data, Ne: number of nodes generating event data, Nn : Number of nodes generating non-real time data, Lc: Transmission time of control data packet, Le: Transmission time of event data packet, Ln: Transmission time of non-real time message at node, [Φ _i , i = 1-M] : Maximum allowable loop delay of control loop i, [ , i = 1-Ne]: average arrival frequency of event data packets at node i, [ , i = 1-Nn]: average arrival frequency of non real-time messages at node i, and [ , i = 1-M]: sampling interval of control loop i, Aℓ above At the beginning of the l th T ₁ slot, N _i is the maximum number of event data that can be generated from all nodes during T ₁ , and α _k is the average of the data originating from the Nc nodes during the T1 interval. The number ν has a step defined as the upper limit number of data generated during the real time interval. Subsequently, the sampling time of the control data To determine the following equations,

It is provided with a step of defining. Then, to determine the first sampling instants of the control data, the following equations,

t ₁ = A ₁ = 0,

for (j = 2; l = 1; j≤Nc, l≤K _M ; j = j + 1, l = ℓ + 1)

t _j = inf [ ]

It is provided with a step of defining. And in order to determine the packet length of the non-real-time data,

" Back side ,

ego Then exit. "

It is provided with a step of defining.

In the CAN bus, the network medium is shared by real time and non real time data, and real time data is subdivided into control and event data. The control data is generated from feedback control loops consisting of a controller and a plant. Sensors and controllers in a control loop periodically generate their data according to a given sampling interval. The sensor and controller data must be transmitted completely within the sampling interval. Event data is usually used for notification of events such as monitoring, diagnostics and alarm information. Event data is generated periodically or sporadically. These data should be transmitted within the time intervals of a particular boundary. Non-real time data consists of programs and data files and database management information. The delay of non real time data is not so important as compared to that of real time data. Thus, non real-time data is transmitted using the extra bandwidth left after real-time data traffic has been allocated. The bandwidth allocation algorithm introduced in the present invention satisfies the real time control and delay requirements of event data. In addition, the redundant bandwidth in which non-real-time data is transmitted is fully utilized to maximize the utilization of the bandwidth of the network medium.

Hereinafter, the abbreviations used in the present invention will be described.

M is the number of control loops interconnected by the CAN bus, Nc is the number of nodes generating control data, Ne is the number of nodes generating event data, Nn is the number of nodes generating non real-time data, and Lc is the control The transmission time of the data packet, Le is the transmission time of the event data packet, Ln is the transmission time of the non-real time message at the node, [ , i = 1-M] is the maximum allowable loop delay of control loop i, [ , i = 1-Ne] is the average arrival frequency of event data packets at node i, [ , i = 1-Nn] is the average frequency of arrival of non-real-time messages at node i, [ , i = 1-M] indicates the sampling interval of the control loop i.

Event data and non real-time messages usually arrive randomly, their typical distribution is unknown. In the present invention, the arrival process of event data and non real-time messages is assumed to have a poison distribution. The queue capacity of control data is limited to one because the most recently generated sensor and controller data must be transmitted. The queue capacity of event and non real-time data is assumed to be large enough so that no data is rejected due to overflow of the transmitter queue when network traffic is stable. Figure 1 shows the basic concept of a bandwidth allocation algorithm in the CAN bus. The bandwidth of the CAN bus is divided by the time slot of the macro cycle equal to T1 which is the minimum of the sampling times between the M control loops. The macro cycle is a real-time (section ( ) And non-real-time segments ( It is divided back into). During the real time period, control data is transmitted using the window scheduling algorithm presented by the applicant. Non-real time data is transmitted during the non-real time interval. Event data is sent with the highest priority because they must be sent as soon as possible.

In order to apply a bandwidth allocation algorithm to the CAN bus, it is necessary to obtain a method of dividing the network bandwidth into real time and non real time intervals. In the CAN protocol, this can be achieved simply by marking the appropriate value in the identifier field of the CAN frame. In the case of the event data, the first two bits of the 11-bit identifier field are set to "00". The first two bits of control and non real-time data are set to "01" and "10", respectively. Since "0" (dominant bit) in the identifier field prevails over "1" (tenth bit), event data has priority over control data, and control data has priority over non-real-time data in transmission. Have This means that non-real time data cannot be transmitted until the end of the real time interval. During a non-real time interval, if a node generates control data, the network bandwidth automatically returns to the real time interval.

Control data is transmitted by the window scheduling algorithm during the real time period. In a window scheduling algorithm, Nc nodes dynamically change the windows. Share it. The window is a time slot whose length is equal to Lc. In order to fit the window scheduling algorithm, the number of data generated during the real time interval should not exceed (ν). In order to limit the number of data sampled (or generated) into (ν), the window scheduling algorithm schedules the beginning of the sampling instant and the sampling period of each control data. Non-real-time data transmitted through the non-real time interval should not infringe the real time interval. Thus, the packet transmission time of the non-real time data needs to be adjusted as necessary.

The basic concept of the window scheduling algorithm is briefly summarized below.

The CAN bus is assumed to have M control loops. Each control loop is assumed to have two data transmission nodes, a sensor and a controller. In the same control loop, the sensor and controller nodes sample their data with one and the same sampling rate. The loop delay is defined from the moment the sensor node samples the sensor data until the controller data generated using the tagged sensor data reaches the actuator node completely. section Consider a control loop i that samples their data with. The loop delay is the delay from the sensor to the controller ( And the network induced delay from the controller to the actuator ( ) Changes with time, too. simply( )Wow ( ) To their minimum upper bounds is not a solution for the design of delay systems that change over time. Let k be the kth instant when the actuator command reaches the plant. ( ) Is time varying (varies with time), so , ) It is also time-varying. The effect of time varying delay on the control system is rather than the amount of time varying delay itself. , Affected by). Above section [ , ] Is the minimum value (min ) Occurs at the kth arrival, and subsequently at the (k + 1) th arrival, the maximum value (max ) When the interval ( , ) Is the maximum. ( , Minimum caps + (sup min )to be. In addition to the controller-actuator delay, it is necessary to consider the sensor-controller delay. The minimum upper limit of to be. So, the minimum upper bound of loop delay is 2 + (sup min )to be. In the window scheduling algorithm, each control data item is a macro cycle. Is transmitted within. So, sup Is And min Is Lc, and the packet transmission time of the control data is to be. In order to meet the performance requirements of the control system, the sampling time of the control loop i must be determined to satisfy the following conditions:

2 + ( - ) ≤ (One)

The maximum allowable loop delay Φ (i = 1-M) of each control loop is assumed to be known in advance. Φ _i can be obtained from the nature of the control loop and from the modified constant version of the time-varying network delay. The elements of the vectors T and Φ are each i th control loop It consists of a sampling interval of (i = 1-M) and a maximum allowable loop delay Φ _i -M.

T = [ , , , .... ] (2)

Φ = [ , , , ... (3)

The vectors T and Φ are defined in increasing order. (In other words, , ) In FIG. 1, and Are the minimum and maximum sampling intervals between the M control loops, respectively.

In the real time interval, the window scheduling algorithm Determine (i = 1-M). Counting Time-varying Characteristics of Network Induced Delay in Control Data In advance = ( + Lc) / 3 is determined from the above equation (1). Sampling time of the rest of the control loops (i = 2-M) is determined so that the number of data generated during the real time interval does not exceed v. As an example, Let's say the average number of data generated from Nc nodes.

K = (4)

Where ν = [ ], Ki = Of to be. And sampling intervals , i = 1 to N are integer multiples of each other (The largest sampling time Of macro cycle All data generated during any section of the It can be accommodated by the windows provided during the same period of. In order to satisfy the performance requirements of the control system, Must satisfy the above conditions given in equation (1). so Is to be determined by the following equation (5).

(5)

Where y = <x> is the power of 2 closest to x but not exceeding , ∈ {0,1,2 ..}. This ensures that the sampling intervals of the M control loops are integer multiples of each other. In Figure 1, (ℓ = 1-M) My ℓ th Is the first moment of the slot. In the window scheduling algorithm, The number of data generated at must not exceed v. In order to ensure this requirement, the window scheduling algorithm performs a sampling instant of each control data item as well as the sampling interval. Schedule until the start of (j = 1-Nc). from j = 1-Nc, Starts at l = 1-M Is assigned to. If the number of allocations in exceeds ν, then the corresponding nodes The next moment ( ) Should be shifted to so, Is determined as follows.

= inf

j = 1-Nc, 1 = 1- Of (6)

here, ( ) Contains the sampling at node j The number of sampled data in. The window scheduling algorithm requires the clocks of the nodes in the common medium to be synchronized. As shown in FIG. 1, undergarment The remainder of is allocated for the transmission of non-real-time data. In the bandwidth allocation algorithm, control data has priority over non-real time data. So, if a node starts transmitting control data, the non-real time interval Will stop automatically. However, non-real-time data items are instantaneous when control data is sampled. If is being transmitted, the transmission of the non-real time data should be completed. Consider the extreme case when the non-real time interval in which only one non-real time data packet can be transmitted is reduced. If the non-real time data packet Ln - If exceeds silver Invades up. So, if the transmission time of non-real-time data is longer than the maximum packet transmission time (the time for transmitting the frame of the maximum size consisting of 130 bits including the worst stuffing bits and interframe space), Ln is expressed by Equation (7). It should be restricted to

Ln = - (7)

For the traffic of the CAN bus system to be stabilized, the frequency of data arrival should be limited to such an extent that it does not exceed the network capacity. The stabilization conditions of the event data and the non-real-time data can be obtained by slightly modifying the results from Ibe and Cheng, who presented the stabilization conditions using the concept of station backlog. In this study, the frequency of arrival of event data packets and non-real-time data is assumed to have a poison distribution. Event data is sent with event data having the highest priority and their transmission is not affected by the transmission of control and non-real time data. Consider the average number of event data arriving at node i over a very long interval of t. For a queuing system to stabilize, the time required for all of those data is on average less than t. The average number of data packets arriving during t is Because of this, the time required for transmitting all of these data packets at node i is ( Le is. Now consider the time required to send event data from other nodes during t. Node j ( ), The average number of data packets transmitted during t Or if If the average number of data packets reached to node j is to be. Thus, the time required to transmit event data from other nodes during t is Becomes For the event data queue at node i for stabilization, all data arriving within t must be transmitted within time t.

(8)

In order to stabilize the network system, all transmitter queues in the network system must be stable. If the queue of the transmitter with the best arrival frequency of event data is stable, all transmitter queues of event data in the network system must be stable. Thus, the stability condition of the event data is as shown in Equation (9) below.

(9)

Non-real-time messages are usually broken up into multiple packets and inserted into the transmitter queue. These packets are sent one by one and combined again at the destination node. The arrival frequency of non real-time data on node i That is Divided into packets, the average number of non-real-time data packets generated per unit time at node i is Becomes Non-real-time data packets are sent with the lowest priority, which is affected by the transmission of event and control data. Consider the average number of non-real time data packets arriving at node i over a very long period of time t. The time required to transmit all these data for the stabilization of the queuing system should average less than t. The average number of data packets arriving in t is Since the time required to send all these data packets is Becomes Consider the time required to transmit non-real-time data from other nodes during time t. Node j ( ), If If, data packets Is sent and otherwise data packets between arrivals at node j Is sent. Thus, the time required to transmit non-real-time data from other nodes during t is Become. In addition, it is necessary to consider the time required for transmitting control data during t. macro cycles during t The number of (t / )to be. The average number of control data generated during to be. Time frame On average to transmit control data during Is taken, so the time required to transfer control data during t to be. Finally, consider the time required to transmit event data during t. The arrival frequency of event data at node k If, the number of event data transmitted during t Becomes So, the time required to send event data during t Becomes In order for the non-real-time data queue to stabilize at node i, all data packets arriving during t must be transmitted in less than t time.

10

In order to stabilize a network system, all transmitter queues in the network system must be stabilized. If the transmitter queue with the non-real time data of the best arrival frequency is stabilized, all transmitter queues in the network system are stabilized. Thus, the stability condition of the network system is given by the following expression (11).

(11)

Consider the stability conditions of the control data. As shown in FIG. 1, the event data may be transmitted during a real time interval, Should not exceed In practical manufacturing automation and process control systems, event data does not occur so frequently compared to control data. Probability of generating n event data during to be.

For example, When = 10ms and The probability that 10 or less event data are generated is 7.89 × 10 ⁻⁵ . In this invention, The maximum number of event data that can be generated from all nodes during Occurred during Preset so that the probability of the following event data is sufficiently small Limited to if , Some control data It will not be sent within and will be rejected from the transmitter queue due to overflow of the queue capacity. Thus, the stochastic stability condition of the control data is given by the following equation (12).

(12)

Since the stability conditions of the event data given in equation (9) are included in equation (11), the CAN bus containing event, control, and non-real-time data is subject to the stability conditions given in equations (11) (12). It must be satisfied at the same time. The aforementioned bandwidth allocation in the CAN protocol is summarized below.

Above Are given values.

First, the sampling time Ti of the control data is determined as the first step.

Let's do it.

ego, Let's do it.

if If the control data traffic is overloaded, M or Ne is reduced and the first step is performed again.

If νL _c + n _i L _e ≤ T ₁ is satisfied, T _i = k _i T _i .

Then, as a second step, determine the first sampling instants t _j of the control data.

Let's do it. At this time, about,

to be.

As a third step, the packet length of the non-real time data is determined.

if, If then to be.

Let's do it.

if In this case, since event or non-real-time data traffic is overloaded, Ne or Nn is reduced and the third step is performed again.

if If the condition is met, the algorithm is terminated.

The bandwidth allocation algorithm in the CAN protocol is verified using simulation experiments. The network system is classified as a discrete event system in which the state of the system changes with the occurrence of an event. Thus, the CAN simulation model is developed using SIMEN / ARENA, a development tool for discrete event systems. In the simulation model, the CAN bus is assumed to have five control loops (M = 5). Each control loop has three nodes: a sensor, a controller, and an actuator. 2 is a table showing the shape of each node in the simulation model.

Node groups (1, 2, 11) (3, 4, 12) (5, 6, 13) (7, 8, 14) and (9, 10, 15) are the control loops (1, 2, 3,4,5). In Fig. 2, data types 1, 2 and 3 represent event, control and non-real time data, respectively. Sensor nodes 2, 4, 6, 8, and 10 have two transmitter queues for control and event data. Controller nodes 1, 3, 5, 7, and 9 also do not have two transmitter queues for control and non-real time data. The CAN bus has 15 nodes and 20 transmitter queues. The number of control, event, and non-real time queues are 10, 5, and 5 (Nc = 10, Ne = 5, Nn = 5), respectively. In FIG. 2, the priority of the frame identifier is allowed in the order of event data, control data and non-real time data. For control loops, control loop 5 is set to the highest priority and control loop 1 is set to the lowest priority. The capacity of the control data transmitter queue is limited to one and that of the event and non-real time data transmitter queues is large enough that no data is saturated due to overflow of the transmitter queue. At this time, the data frequency B of the CAN bus is set to 125 KB / s.

Sensor data and control signals are generated periodically in each control loop. The length of control data is assumed to be 8 bytes, which is the maximum length of the data packet. That is, the total length of the control data is 130 bits including overhead, worst case stuffing bits, and frame interspace (3 bits), and the packet transmission time Lc is 1.04 ms. The sensor node in each control loop sends monitoring data to the supervisory controller node. The mutual arrival time of the monitoring data is 10 ms (with an exponential distribution on average, which is intentionally added high traffic load). )to be. The length of event data is assumed to be 2 bytes because event data typically carries a limited amount of information. The total length of event data is 73 bits, including overhead, worst case stuffing bits and interframe space, and packet transmission time Le is 0.584 ms. It is assumed that the controller node in each control loop sends file data to the supervisory controller node. The length of non-real-time data is assumed to be 30 KB. In the real case, file data is delivered from higher layers and packetized before transmission. In the present invention, the arrival of the non-real time data packet is 17.065 ms (where the mutual arrival time has an exponential distribution on average). Modeled to The supervisory controller node is not assumed to generate any message, and the actuator node in each control loop does not just receive data and transmit. As an example, the maximum allowable delay of event data is assumed to be 10 ms. The maximum loop delay of the five control loops is given as [Φ ₁ , Φ ₂ , Φ ₃ , Φ ₄ , Φ ₅ ] = [29ms, 60ms, 120ms, 240ms, 480ms]. From step 2 of the algorithm given above, Φ ₁ and T ₁ are set to 29ms and 10ms, respectively, and [k ₁ , k ₂ , k ₃ , k ₄ , k ₅ ] = [1,2,4,8,16]. The average number of control data sampled from five control loops during T ₁ is = 3.875 and the number of windows in the real time interval is = 4. For example, the probability that an abnormal occurrence of n _i event data in T ₁ should be <5 × 10 −2, and n _i is required to be determined to be 10. Because of this, the traffic of the control data is stochastically stable. The vector of sampling time in the five control loops is [T ₁ , T ₂ , T ₃ , T ₄ , T ₅ ] = [10 ms, 20 ms, 40 ms, 80 ms, 160 ms]. From step 2, the first sampling instant of control data from nodes 1-10 is [t ₁ , t ₂ , t ₃ , t ₄ , t ₅ , t ₆ , t ₇ , t ₈ , t ₉ , t ₁₀ ] = [0ms, 0ms, 0ms, 0ms, 10ms, 10ms, 30ms, 30ms, 70ms, 70ms]. From step 3 Is greater than the transmission time of the maximum packet length (130 bits) of the CAN protocol. Thus, the transmission time of the non real-time data packet is set to Ln = 1.04 ms.

Because of this, the traffic of the network system is stable, and the network bandwidth is almost fully utilized.

3 shows the results obtained from the simulation model.

In Fig. 3, the algorithms are the results when the CAN bus adopts the bandwidth algorithm already described. The traffic condition of the non-algorithm is the same as that of the algorithm as given in Table 1. However, the non-algorithm does not take into account the first sampling instant of the control data (ie t _i = 0 ms, i = 1-Nc). So, it does not follow the bandwidth allocation algorithm. The rejected data frequency in Table 2 indicates the frequency of the rejected data number relative to the generated data number. Here, the rejected data means data rejected in the transmitter queue due to overflow of the queue capacity.

3 shows that the average and maximum data waits of the event data are nearly equivalent in both cases because the event data is sent with the highest priority. The maximum data wait time of event data is 3.93ms less than the maximum allowable data wait 10ms. In the case of control data, the maximum data wait in the algorithm is 9.50 ms. It does not exceed the minimum sampling interval T ₁ = 10 ms and no data is rejected. However, in the case of non-algorithms, the maximum data wait for control data is 14.81 ms (> 10 ms) and 19 (474 or more) data is rejected in the transmitter queue. The data wait of non-real-time data has no significant significance in the performance analysis of the CAN bus. No non-real-time data is rejected because the traffic conditions are stable. Figure 3 limits the delay of event and control data to a specified limit in such a way that the bandwidth allocation algorithm introduced in the present invention increases the delay of non-real-time data that does not affect the performance of application systems.

Figure 4 shows that the traffic conditions are the same as in the previous case as given in Figure 2 but the data frequency is reduced to 100 KB / s.

In this case, the stability conditions given in the algorithm And the traffic load exceeds the capacity of the network bandwidth. 3 shows that the effect of increasing traffic load is mainly exerted on the non-real-time data with the lowest priority classification. Because the network system is unstable, non-real-time data is rejected in the transmitter queue (in this simulation analysis, the queue capacity of non-real-time data is set to 133). However, the maximum data wait for event data is still the maximum allowed data wait. Less than 10ms, no event data is rejected. In the case where the bandwidth allocation algorithm is applied, the maximum data wait of the control data is 9.99 ms which is still less than the maximum allowable delay of 10 ms, and no control data is rejected. Thus, the bandwidth allocation algorithm satisfies the performance requirements of event and control data even if the traffic condition is overloaded to some extent.

The effect of rejection of control data on the performance of the feedback control system is illustrated below. Control loop 1 consists of nodes (1, 2). The transfer function of the plant at node 2 is chosen by the equation

Similarity of the transmission function of the digital controller at node 1 is expressed by the following equation.

Where s is a Laplace Transformation Variable. 5 (A) and (B) show the performance comparison of algorithms and non-algorithms when B = 125 KB / s. As shown in Fig. 3, various control data are rejected in case of non-algorithm. This data rejection causes jitter in the controller signal, which causes high frequency noise in the actuator command that leads to excess consumption of the actuator. The response of the plant output is also delayed and its performance is also degraded. Here, the stability conditions of the network system given in Equation (11) can be confirmed using simulation experiments. 6 shows the change in the number of waiting messages in the non-real-time data transmitter queue for different arrival frequencies of the non-real-time data to verify the stabilization conditions of the non-real-time data given in equation (11). Traffic loads of event and control data are given above example and as B = 125 KB / s. 6 is a condition that satisfies equation (11). , The number of waiting messages is ultimately stable. However, the condition that does not satisfy equation (11) When, the number of waiting messages is increased as time passes, and the CAN bus system becomes unstable.

The CAN is a network protocol that supports field devices in a distributed control system. Data generated from field devices is classified into three categories: control, event and non-real time. Excess data latency of event and control data may reduce the performance of application systems interconnected with the CAN bus. The present invention introduces a bandwidth allocation algorithm within the CAN bus. The bandwidth allocation algorithm not only takes full advantage of the bandwidth of the CAN bus but also satisfies the performance requirements of real-time application systems. The bandwidth allocation algorithm is verified using a simulation model that integrates the discrete-event model of the CAN protocol and the continuous time model of the feedback control system. The bandwidth allocation algorithm introduced in the present invention can be effectively used in the design and construction phase of distributed real-time control and automation systems using the CAN protocol.

Claims (14)

It has a plurality of nodes and a plurality of control loops that share this CAN bus with a control area network bus (CAN), with each control loop allocating bandwidth within a CAN protocol including at least one data transmission node, a sensor and a controller. In the way, Dividing the bandwidth of the CAN bus into time slots of a macro cycle equal to a minimum time of sampling times between the control loops; Marking an appropriate value in the identifier fields of those data so as to have a transmission priority in order of event data which is real time data, control data which is real time data, and non real time data; Subdividing the macro cycle into real-time and non-real-time sections; And And transmitting the event data and the control data in this order during the real time period, and transmitting the non-real time data during the non-real time period. The method of claim 1, further comprising the step of automatically returning the network bandwidth to the real-time interval when at least one node generates control data in the non-real-time interval. . 2. The method of claim 1, wherein the first two bits of the identifier field of the event data are "00", the first two bits of the identifier field of the control data are "01", and the first of the identifier field of the non-real-time data. A method of allocating bandwidth within a CAN protocol characterized by two bits being set to "10". The method of claim 1, further comprising: first setting a plurality of windows to transmit the control data by a window scheduling algorithm during the real-time period; Setting a length of each window to a time slot having a length equal to a packet transmission time of the control data; Sharing the windows by nodes generating the control data; And And scheduling a start of a sampling time and a sampling interval of each control data so as to limit the number of the control data to a predetermined number or less for the window scheduling algorithm. . 5. The method of claim 4, wherein the sampling time T _i of the control loop i of the plurality of control loops is Φ ₁ is the maximum allowable loop delay time in the control loop i, T _l is the minimum value of the sampling times of the control loops , L _c is the following condition when the packet transmission time of the control data, 2T _i + (T _l -L _c ) ≤ Φ _i A method of allocating bandwidth within the CAN protocol characterized by the fact that The method of claim 4, wherein M is the number of the control loops, a sampling interval of control loop i of the plurality of control loops is T _i , k _i is T _i / T _l , N _c is the number of nodes generating control data. number, α _k is the number average of the data generated from the N _c nodes, L _c is the transmission time of the control data packets, Φ _i is at the maximum allowable loop delay of the control loop i, the sampling interval T _i is expression, A method for allocating bandwidth within a CAN protocol, characterized in that obtained by performing the following steps. The method of claim 4, wherein ν is a maximum allowable number of control data generated during a real time interval, M is the number of the control loops, T _l is the minimum sampling interval between the control loops, and T _M is the interval between the control loops. the maximum sampling interval, each control data t _j sampling instant of the _{(j = 1-N c)} , a l is the T _M I l the first moment of the first said T ₁ slots, U _j (a _l) is the node j The number of data sampled in A _l , including sampling at, N _c is the number of nodes generating control data, arbitrary node j is 1-N _c , and l is 1-T _M / T _l , The sampling instant t _j of the control data is expressed by tj = inf [A _l ≥ A _l-1 ; U _j (A _l ) ≤ ν]. A method of allocating bandwidth within a CAN protocol. The method of claim 1, wherein L _n is a transmission time of a non-real time message at the nodes, τ _n is a transmission interval of non-real-time data, τ _γ is a transmission interval of real-time data, T _l is the minimum between the control loops When the sampling period, the transmission time L _n of the non-real-time message should be limited to L _n = T _l -τ _γ method for allocating bandwidth in the CAN protocol. The method of claim 1, wherein for node j, the number of event data packets transmitted When N _e is the number of nodes generating event data and L _e is the transmission time of an event data packet, in order to stabilize the network system, the stability condition of the event data is expressed by: Allocating bandwidth within the CAN protocol, characterized in that The method of claim 1, wherein T ₁ is a minimum value of sampling times of the control loops, α _k is an average number of control data generated from the N _c nodes, N _e is a number of nodes generating event data, N _n is the number of nodes generating non-real time data, L _c is the transmission time of the control data packet, L _e is the transmission time of the event data packet, L _n is the transmission time of the non real-time data at the node, event data at node k Frequency of arrival And the average number of non-real time data packets generated per unit time at node j is When the stability condition of the network system is A method of allocating bandwidth within a CAN protocol characterized in that it must satisfy The method of claim 1, wherein T _l is a minimum value of sampling times of the control loops, L _c is a transmission time of the control data packet, L _e is a transmission time of an event data packet, and n _i is all nodes during the T _l . When the maximum number of event data that can be generated from the field, ν is the upper limit of the data generated during the real-time interval, the stability condition of the control data is νL _c + n _i L _e ≤ T _l A method of allocating bandwidth within a CAN protocol characterized by that A method of allocating bandwidth within a CAN protocol having a CAN bus and a plurality of nodes and a plurality of control loops sharing the CAN bus, each control loop comprising at least one data transmission node, a sensor and a controller. (a) M: number of control loops interconnected by CAN bus, N _c : number of nodes generating control data, N _e : number of nodes generating event data, N _n : node generating non real-time data Number of fields, L _c : transmission time of control data packet, L _e : transmission time of event data packet, L _n : transmission time of non real-time message at node, [Φ _i , i = 1-M]: of control loop i Allowable loop delay, [ , i = 1-N _e ]: average arrival frequency of event data packets at node i, [ , i = 1-N _n ]: average arrival frequency of non real-time messages at node i, and [T _i , i = 1-M]: sampling interval of control loop i, A _l is the first said T in said T _M the first moment of _l slot, n _i is in the T _l all nodes the maximum number of event data that may be generated from, α _k is the n _c nodes the number average of the data generated from the, ν is a real-time interval The upper limit of the generated data, P _j is the number of packet segments at node j, Is the frequency of arrival of the non-real-time data at node j, defined by; (b) in order to determine the sampling time T _i of the control data, , , If νL _c + n _i L _e <T _l then T _i = K _i T _l Performing a step; (c) in order to determine the first sampling instants of the control data, When t ₁ = A ₁ = 0, t _j = inf [A _l for (j = 2; l = 1; j ≤ N _c , l ≤ K _M ; j = j + 1, l = l + 1) ≥ A _l-1 ; u ^j (A _l ) ≤ ν] Performing a step; (d) in order to determine the packet length of the non-real-time data, If L _n T _1- (νL _c + n _i L _e ) then L _n = T _1- (νL _c + n _i L _e ) Performing a step; And (e) equation, (only, ) And if it satisfies, terminating the bandwidth in the CAN protocol. 15. The CAN as claimed in claim 14, further comprising: if the traffic of the control data is overloaded in step (b), reducing the one of M and N _e and performing step (b) again. How to allocate bandwidth within the protocol. 15. The method of claim 14, wherein if any one of the traffic of the event data and the non-real-time data is overloaded in the step (e), reducing any one of the N _e and N _n and performing the step (d) again. Further comprising a bandwidth allocation in the CAN protocol.