JP3766603B2 - Cell readout scheduling apparatus and method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention is used for an apparatus (for example, an ATM (Asynchronous Transfer Mode) node apparatus) that multiplexes a plurality of connections on a single line. In particular, the present invention relates to a technique for using an output band required for each connection within a set allowable range of cell transmission interval fluctuation.
[0002]
[Prior art]
In the node equipment constituting an ATM network, when cells input from a plurality of input lines are multiplexed on a single output line, a cell queue corresponding to each input line is prepared as one method for realizing this. Then, a technique is used in which cells are periodically read from the cell queue at a frequency corresponding to the bandwidth required by each input line and the cells are sent to the output line. At this time, as an algorithm for periodically reading cells from each cell queue having different set bandwidths, for example, there is a WRR (Weighted Round Robin) method.
In the WRR method, the cell queue to be read is weighted in proportion to the set bandwidth, and the number of cells corresponding to this weight are read from each cell queue during one cycle. With this mechanism, any cell queue can be expected to ensure a fair output bandwidth according to the set bandwidth.
[0003]
However, when cell multiplexing is performed by the WRR method, the cell transmission interval of each multiplexed connection depends on the cell reading method from each cell queue within one period. If the cell transmission interval of each connection is to be kept constant, it is necessary to make the cell read timing from the cell queue constant within a period, and also make the cell read interval constant between adjacent periods. is there.
[0004]
However, when the number of connections to be multiplexed is large or the bandwidth occupancy rate of the multiplexed line is high, the cell transmission interval of each connection may be affected by the cell transmission status of other connections. That is, if a plurality of connections try to output a cell at the same timing, blocking occurs in the cell multiplexing unit, and a delay occurs in the cell output timing of the connection that could not be output. Therefore, when multiple connections are multiplexed with a high bandwidth occupancy rate, there is a possibility that the cell transmission interval required by each connection cannot be observed.
[0005]
This becomes a problem when trying to schedule a strict cell transmission interval in order to maintain high line quality in the network. Therefore, in order to solve this problem, an appropriate cell read scheduling means is required.
[0006]
As one of the methods, a method of performing cell readout based on a timing scheduling table prepared in advance for a cell readout schedule within one period for all connections is used.
[0007]
This scheduling table is composed of one time slot period that is a cell transmission timing.
(A) Allocate time slots in proportion to the bandwidth of each connection.
(B) When there are a plurality of time slots to be assigned to each connection within one period, they are arranged at regular intervals without being unevenly distributed within the period.
(C) Even when viewed in a plurality of cycles, the time slots are arranged so that the intervals are constant.
(D) Each time slot period interval has a certain allowable range with respect to an ideal arrangement position calculated from the allocated bandwidth.
Suppose that.
[0008]
The following procedure is an example of a method for creating this scheduling table. First, an array of (number of timeslots) × (number of connections) is prepared, and this is used as a scheduling array. When the initial placement time slot (first time slot) of each connection is determined by a random number, the corresponding position in the array is marked. When the initial arrangement time slot is determined, the time slot indicating the subsequent cell transmission timing is determined based on the ideal interval and is marked in the scheduling array. This is repeated for one cycle. However, since the initial arrangement time slot is determined by a random number, it is not positioned at the beginning of one cycle. Therefore, after the time slot at the final position given as an array, an operation for returning to the time slot at the head position is performed. Thereby, it is possible to assume an arrangement in which time slots for one period are continuous from the initial arrangement time slot. Repeat the same operation for all connections to build a scheduling array.
[0009]
When the scheduling array is constructed, connections to be arranged from the first time slot in one cycle are determined in order. For each time slot, all connections are searched in order, and if there are multiple connections marked in the scheduling array, how much from the ideal interval based on the previous time slot allocation position for each connection. Calculate what it is. The connection with the largest deviation is placed in the time slot. If the shifts are the same size, a connection with a large bandwidth is preferentially arranged. This operation is repeated for all time slots so that there are no time slots that are duplicated in all time slots. The scheduling table can be created by the above operation.
[0010]
[Problems to be solved by the invention]
In the conventional technique, as the number of connections to be handled increases, calculation processing for scheduling becomes more complicated, and it takes much time to create a scheduling table. No matter how good the scheduling table is, it is not practical if it requires a huge amount of time to generate it. Therefore, it is necessary to have means for enabling the scheduling table to be generated at a higher speed.
[0011]
In addition, there is a drawback that the expandability is poor, for example, it is difficult to cope with the case where the allowable time slot arrangement position fluctuation value is different for each connection.
[0012]
The present invention has been made in view of such a background, and provides a cell read scheduling apparatus and method capable of generating a scheduling table having characteristics equivalent to or higher than those of a conventional scheduling method at a higher speed than the conventional one. The purpose is to provide. It is an object of the present invention to provide a cell read scheduling apparatus and method that can cope with the introduction of a per-connection time slot arrangement position fluctuation tolerance.
[0013]
[Means for Solving the Problems]
A first aspect of the present invention is a cell read scheduling apparatus, which is characterized by connection information holding means for recording a set of identifier information and bandwidth information of each connection and a specified policy. A connection information sorting means for rearranging a set of connection information recorded in the connection information holding means, a connection identification means for identifying a connection assigned to the time slot arranged for each time slot on the scheduling table, A means for connecting and managing connection identification means for connections to be arranged in time slots in a list for each time slot, and a time slot management means for accepting reservation of cell output timing requested by each connection and arranging the connection identification means And this Means for reserving the first time slot of each connection and the ideal time slot arrangement of each connection starting from this first time slot, and one time slot on the scheduling table. When multiple connections have reserved reservations for the connection identification means, they are distributed to different time slots. Make And a means for reading a cell according to a scheduling result of the scheduling means.
[0014]
That is, in the present invention, in order to reduce the number of times the scheduling array is referenced when creating the scheduling table, the connection identification means that reserves the time slot on the scheduling table is associated with, for example, a pointer indicating the connection identification means for efficient connection. Make a reference. This can reduce the number of times the scheduling array is referenced.
[0015]
Since the connection information for time slot reservation is handled by a list structure connected by a pointer, sliding of the time slot arrangement position is realized by changing the pointer. This makes it possible to execute time slot arrangement processing when a plurality of connections reserve the same time slot at high speed.
[0016]
More specifically, the connection information sorting means includes means for referring to bandwidth information of each connection and sorting in order from a wide bandwidth, and the time slot management means starts from each connection according to the sort order of the sorting means. Means for accepting a reservation of an ideal time slot arrangement for each connection starting from the time slot and the leading time slot, and the means for distributing refers to the acceptance result of the means for accepting and When a plurality of connections for the time slot are reserved for overlapping the connection identification means, the second and subsequent ones of the connection identification means connected in a list are slid to the adjacent time slot, At this time, When this backward time slots other of said connection identification means are arranged, it is desirable to have a means to transfer at the end of the list of the connection identification means is slid from the front to the other of the connection identification means.
[0017]
Further, the connection information holding means includes means for recording together fluctuation tolerance value information of an ideal cell transmission time slot of each connection, and the tolerance value information is set in the connection identification means and the means for distributing the information Refers to the acceptance result of the accepting means, and when a plurality of connections reserve and arrange the connection identifying means for one time slot on the scheduling table, the connection value is set in ascending order. A first step of rearranging the identification means, and a timing at which the second and subsequent connection identification means of the rearranged list are slid to an adjacent rear time slot and the allowable value of the slid connection identification means is slid A second step to reduce by minutes And line, in this case, already when this backward time slots other of said connection identification means is disposed, again, may comprise means for executing said first and second step.
[0018]
That is, an arrangement position fluctuation allowable value is given to the connection information for time slot reservation, a movement range is set for the slide processing on the scheduling table, and management is performed based on the list length of the connection information. As a result, it is possible to set different arrangement position fluctuation tolerances for each connection.
[0019]
The elements of the list are rearranged in ascending order according to the arrangement position fluctuation allowable value of the connection information for time slot reservation. This improves the multiplexing efficiency when the multiple connections to be multiplexed have different arrangement position fluctuation tolerances.
[0020]
A second aspect of the present invention is a cell read scheduling method, which is characterized in that the bandwidth information of each connection is referred to and sorted in descending order from the wide bandwidth, and in accordance with this sort order, the first from each connection Connection identification means for accepting reservations for an ideal time slot arrangement of each connection starting from the time slot and the first time slot, and identifying the connection that reserved the time slot for each time slot on the scheduling table The connection identification means of the connection to be arranged in each time slot is managed by connecting in a list for each time slot, and a plurality of connections overlap the connection identification means for one time slot on the scheduling table. Shi When the reservation is made, the second and subsequent ones of the connection identification means connected in a list for one time slot on the scheduling table are slid to the adjacent rear time slot. When another connection identification unit is arranged in the time slot, the other connection identification unit is moved from the front to the end of the list of connection identification units, and as a result, arranged in each time slot. The number of connection identifying means is limited to 0 or 1, thereby determining the cell read time slot.
[0021]
In addition, when there is room for sliding the second and subsequent connections of the connection identification means ahead of the time slot in which the first of the connection identification means is arranged, one time slot on the scheduling table When a plurality of connections reserve and arrange the connection identification means overlappingly, the second or later connection identification means connected in a list to one time slot on the scheduling table are adjacent to the front or Slide to the rear time slot. At this time, if another connection identification means is already arranged in the front or rear time slot, the list of the connection identification means is slid from the rear or front. Move to the end, and with this result The number of the connection identification means arranged in each time slot Te was limited to 0 or 1, and thereby determine the cell read time slot. Thereby, the freedom degree of time slot arrangement | positioning further improves.
[0022]
In addition, when reserving the first time slot of each connection and the ideal time slot arrangement starting from this first time slot, the time slot fluctuation of the corresponding connection bandwidth information is allowed for the connection identification means to be arranged. When setting the value and distributing the connection identification means connected in the form of a list and arranging them in neighboring time slots, refer to the information on the allowable value of the connection indicated by the connection identification means included in the list, A third step of rearranging the connection identification means so that the permissible value is in ascending order, and sliding the second and subsequent ones of the rearranged list to an adjacent rear time slot and sliding the connection identification means The fourth is to decrease the allowable value by the slide timing. And a step executed, when the connection identification means is located already behind the time slot in this case, again, it is also possible to perform the third and fourth step.
[0023]
Instead of the part handling a single connection as a scheduling unit, a connection group composed of one or more connections can be handled as a scheduling unit.
[0024]
A third aspect of the present invention is a program, which is installed in an information processing apparatus, and is designated with a connection information holding function for recording a set of identifier information and bandwidth information of each connection in the information processing apparatus. A connection information sorting function for rearranging a set of connection information recorded in the connection information holding function based on a policy, and a connection identification function for identifying a connection allocated to each time slot arranged for each time slot on the scheduling table A function for connecting and managing the connection identification function of the connection arranged in each time slot in a list for each time slot, and a time for accepting reservation of cell output timing requested by each connection and arranging the connection identification function slot A function for reserving the first time slot of each connection and the ideal time slot arrangement of each connection starting from this first time slot for the time slot management function, and one function on the scheduling table. A scheduling function including a function of distributing a plurality of connections to a time slot when the connection identification function is reserved and arranged in a different time slot is realized, and a scheduling result of the scheduling function is obtained. Therefore, the cell reading function is realized.
[0025]
The connection information sorting function has a function of referring to bandwidth information of each connection and sorting in order from a wide bandwidth, and the time slot management function includes the first time slot from each connection and the time slot according to the sort order of the sorting function. A function of accepting reservations of ideal time slot arrangement of each connection starting from the first time slot, and the function of distributing refers to the acceptance result of the means for accepting, with respect to one time slot on the scheduling table When a plurality of connections are reserved for overlapping the connection identification function, the second and subsequent ones of the connection identification functions connected in the form of a list are slid to the adjacent rear time slot. Already behind this tie When the other of the connection identification function in the slot is arranged, it is desirable to realize a function to transfer the end of the list of the connection identification features this other said connection identification feature is slid from the front.
[0026]
The connection information holding function includes a function to record fluctuation tolerance value information of ideal cell transmission time slots of each connection together, the tolerance value information is set in the connection identification function, and the function to distribute is The connection identification function so that the allowable value is in ascending order when a plurality of connections reserve the connection identification function for one time slot on the scheduling table with reference to the reception result of the reception function. And the second and subsequent connection identification functions of the rearranged list are slid to the adjacent rear time slot, and the allowable values of the slid connection identification functions are slid by the timing of sliding. Perform a second step to reduce and At this time, already when another said connection identification function to the rear of the time slot is located, again, it is desirable to realize a function to execute the first and second step.
[0027]
Alternatively, the third aspect of the present invention is a program, which is installed in an information processing device, and in the information processing device, the bandwidth information of each connection is referred to and sorted in order from the wide bandwidth, and according to this sort order. Receiving reservations for the first time slot from each connection and the ideal time slot arrangement of each connection starting from this first time slot, each time slot on the scheduling table identifies the connection that reserved that time slot And a connection identification function for connections arranged in each time slot is managed by connecting in a list for each time slot, and a plurality of connections are associated with one time slot on the scheduling table. Co When a duplicated function identification function is reserved, the second and subsequent connection identification functions connected in a list for one time slot on the scheduling table are slid to the adjacent rear time slot. When another connection identification function is already arranged in this rear time slot, the other connection identification function is moved to the end of the list of the connection identification functions slid from the front. As described above, the number of connection identification functions arranged in each time slot is limited to 0 or 1, thereby executing a procedure for determining a cell read time slot.
[0028]
When there is room to slide the second and subsequent connections of the connection identification function ahead of the time slot in which the first of the connection identification functions is arranged, a plurality of time slots on the scheduling table When a connection is reserved for overlapping the above-described connection identification function, the second or later connection identification function connected in a list with respect to one time slot on the scheduling table is adjacent to the forward or backward time. If the other connection identification function is already arranged in this time slot at the front or rear at this time, it is slid from the rear or the front to the end of the list of the connection identification functions. As a result of this The number of the connection identification feature disposed in the slot is limited to 0 or 1, which is executed by the steps of determining a cell read time slot is desirable.
[0029]
When reserving the first time slot of each connection and the ideal time slot arrangement starting from this first time slot, the time slot fluctuation allowable value of the connection bandwidth information corresponding to the connection identification function to be arranged is set. When setting and distributing the connection identification functions connected in a list form and arranging them in neighboring time slots, refer to the information on the allowable value of the connection indicated by the connection identification function included in the list, and A third step of rearranging the connection identification functions so that the values are in ascending order; and the second and subsequent ones of the rearranged list are slid to an adjacent rear time slot, and the slide of the connection identification function A fourth step that reduces the tolerance by the slide timing. Run the flop, when the connection identification function is arranged already behind the time slot in this case, again, it is desirable to perform the procedure for performing the third and fourth step.
[0030]
Instead of the part handling a single connection as a scheduling unit, a connection group composed of one or more connections can be handled as a scheduling unit.
[0031]
DETAILED DESCRIPTION OF THE INVENTION
A cell read scheduling apparatus and method according to an embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a block diagram of a cell read scheduling apparatus according to an embodiment of the present invention. FIG. 2 is a flowchart showing the procedure of the cell read scheduling method according to the embodiment of the present invention. FIG. 3 is a conceptual diagram of scheduling in one period of the scheduling table. FIG. 4 is a flowchart showing a procedure when a scheduling table is created a plurality of times in order to create a scheduling table with little deviation from the ideal position.
[0032]
A first aspect of the present invention is a cell read scheduling apparatus, which is characterized in that, as shown in FIG. 1, connection information holding means 1 for recording a set of identifier information and bandwidth information of each connection. And a connection information sorting means 2 for rearranging the set of connection information recorded in the connection information holding means 1 based on the designated policy, and a connection allocated to the time slot arranged for each time slot on the scheduling table. A connection identification means 3 for identifying the connection, a list management means 4 for managing the connection identification means 3 of the connection arranged in each time slot in a list for each time slot, and reservation of cell output timing required by each connection Accepts and places connection identification means 3 Time slot management means 5 and the time slot management means 5 for the first time slot of each connection (initially arranged time slot) and the ideal time slot arrangement of each connection starting from this first time slot. The initial arrangement time slot reservation means 6 and the ideal arrangement time slot reservation means 7 to reserve, and when a plurality of connections reserve the connection identification means 3 for one time slot on the scheduling table, this is set to different time slots. There is a scheduling means 9 including an arrangement time slot distribution means 8 arranged in a distributed manner, and a cell reading means 10 for reading cells according to the scheduling result of the scheduling means 9.
[0033]
More specifically, the connection information sorting unit 2 refers to the bandwidth information of each connection and sorts the connections in descending order, and the time slot management unit 5 sorts the first time from each connection according to the sorting order of the connection information sorting unit 2. Reservation of an ideal time slot arrangement of each connection starting from the slot and the first time slot is accepted (processing of S1 and S2 in FIG. 2), and the arrangement time slot distribution means 8 performs scheduling by referring to the acceptance result. When a plurality of connections reserve and arrange the connection identification means 3 for one time slot on the table, the second and subsequent ones of the connection identification means 3 connected in a list form are connected to the rear side. Slide it into the time slot (S in Figure 2) , S4) At this time, when another connection identification means 3 is already arranged in the rear time slot, the list of connection identification means 3 in which the other connection identification means 3 is slid from the front is displayed. Move to the end.
[0034]
Alternatively, as the processing of step S5 in FIG. 4, the connection information holding unit 1 records the fluctuation allowable value information of the ideal cell transmission time slot of each connection and sets the allowable value information in the connection identifying unit 3. The arrangement time slot distribution means 8 refers to the reception result of the time slot management means 5 and when a plurality of connections reserve and arrange the connection identification means 3 for one time slot on the scheduling table, The first step of rearranging the connection identification means 3 so that the allowable value is in ascending order, and the second and subsequent connection identification means 3 of the rearranged list are slid to the adjacent rear time slot, and the slide connection identification is made. By the timing of sliding the allowable value of means 3 Run a second step to lack, at this time, already when the other connection identification means 3 to the rear of the time slot are arranged, again, performing the first and second step. As a result, the scheduling table with the smallest maximum arrangement position fluctuation is adopted.
[0035]
A second aspect of the present invention is a cell read scheduling method applied to the cell read scheduling apparatus shown in FIG. 1, and the feature of the present invention is that the band information of each connection is referred to in order from the wide band. In accordance with this sort order, the first time slot from each connection and the reservation of the ideal time slot arrangement of each connection starting from this first time slot are accepted, and each time slot on the scheduling table corresponds to the corresponding time slot A connection identification means 3 for identifying a connection that reserves the connection, and the connection identification means 3 of the connection arranged in each time slot is connected and managed in a list for each time slot, and one time on the scheduling table Against the slot When a plurality of connections reserve the connection identification means 3 redundantly, the second and subsequent connections among the connection identification means 3 connected in a list for one time slot on the scheduling table are adjacent to the rear time. At this time, when another connection identification means 3 is already arranged in the rear time slot, the other connection identification means 3 is slid from the front to the end of the list of connection identification means 3 As a result, the number of connection identifying means 3 arranged in each time slot is limited to 0 or 1, thereby determining a cell read time slot.
[0036]
When there is room for sliding the second and subsequent ones of the connection identification means 3 ahead of the time slot where the first of the connection identification means 3 is arranged, a plurality of time slots on the scheduling table When a connection is reserved by overlapping the connection identification means 3, a second or later connection identification means 3 connected in a list for one time slot on the scheduling table is adjacent to the front or rear. When another connection identification means 3 is already placed in this time slot at the front or rear at this time, the last of the list of connection identification means 3 that has been slid from the rear or the front. As a result, each time slot The number of arranged connection identification means 3 to be limited to 0 or 1, thereby determining the cell read time slot.
[0037]
Further, when reserving the first time slot of each connection and the ideal time slot arrangement starting from this first time slot, the connection identification means 3 to be arranged can accept time slot fluctuations in the corresponding connection bandwidth information. When setting the value and distributing the connection identification means 3 connected in the form of a list and arranging them in neighboring time slots, refer to the information on the allowable value of the connection indicated by the connection identification means 3 included in the list, A third step of rearranging the connection identification means 3 so that the permissible values are in ascending order, and the second and subsequent ones of the rearranged list are slid to the adjacent rear time slot, and the slid connection identification means 3 The fourth step is to decrease the allowable value by the slide timing. Run the door, when the already behind the time slot at this time connection identification means 3 is disposed, again, performing the third and fourth step.
[0038]
Instead of the part handling a single connection as a scheduling unit, a connection group composed of one or more connections can be handled as a scheduling unit.
[0039]
A third aspect of the present invention is a program, and the cell read scheduling apparatus of the present invention installs the program of the present invention in a personal computer apparatus or the like as an information processing apparatus, thereby This can be realized by using a device corresponding to the cell readout scheduling device of the present invention. With this personal computer device or the like, the cell readout scheduling method of the present invention can be executed.
[0040]
That is, the program of the present invention has a connection information holding function for recording a set of identifier information and band information of each connection in a personal computer device or the like by installing it in a personal computer device or the like as an information processing device. A connection information sorting function for rearranging a set of connection information recorded in the connection information holding function based on a specified policy, and a connection assigned to the time slot arranged for each time slot on the scheduling table A connection identification function that connects to each time slot, a connection management function that manages the connection identification function in a list for each time slot, and reservation of cell output timing required by each connection The time slot management function that arranges the connection identification function, the first time slot of each connection, and the ideal time slot arrangement of each connection starting from this first time slot are reserved for this time slot management function Realizes a scheduling function that includes a function and a function that distributes and arranges the connection identification functions in different time slots when a plurality of connections are reserved for one time slot on the scheduling table. The cell reading function is realized according to the scheduling result of the scheduling function.
[0041]
The connection information sorting function has a function of referring to bandwidth information of each connection and sorting in order from a wide bandwidth, and the time slot management function includes the first time slot from each connection and the time slot according to the sort order of the sorting function. A function of accepting reservations of ideal time slot arrangement of each connection starting from the first time slot, and the function of distributing refers to the acceptance result of the means for accepting, with respect to one time slot on the scheduling table When a plurality of connections are reserved for overlapping the connection identification function, the second and subsequent ones of the connection identification functions connected in the form of a list are slid to the adjacent rear time slot. Already behind this tie When the other of the connection identification function in the slot is arranged can be implemented by a function to transfer the end of the list of the connection identification features this other said connection identification feature is slid from the front.
[0042]
The connection information holding function includes a function to record fluctuation tolerance value information of ideal cell transmission time slots of each connection together, the tolerance value information is set in the connection identification function, and the function to distribute is The connection identification function so that the allowable value is in ascending order when a plurality of connections reserve the connection identification function for one time slot on the scheduling table with reference to the reception result of the reception function. And the second and subsequent connection identification functions of the rearranged list are slid to the adjacent rear time slot, and the allowable values of the slid connection identification functions are slid by the timing of sliding. Perform a second step to reduce and At this time, already when the other of the connection identification feature behind the time slot is located can be re-implemented a function to execute the first and second step.
[0043]
According to the program of the present invention, by making the personal computer device a device corresponding to the cell readout scheduling device of the present invention, the personal computer device is sorted in order from the wide bandwidth with reference to the bandwidth information of each connection, In accordance with this sort order, the first time slot from each connection and the reservation of the ideal time slot arrangement of each connection starting from this first time slot are accepted, and the corresponding time slot is reserved for each time slot on the scheduling table. A connection identification function for identifying the connection is arranged, and the connection identification function of the connection arranged in each time slot is connected and managed in a list form for each time slot, and one on the scheduling table When a plurality of connections reserve the same connection identification function for a time slot, the second and subsequent connection identification functions connected in a list for one time slot on the scheduling table are adjacent. The other connection identification function is slid from the front when another connection identification function is already arranged in this rear time slot. As a result, the number of connection identification functions arranged in each time slot is limited to 0 or 1, thereby executing a procedure for determining a cell read time slot.
[0044]
When there is room to slide the second and subsequent connections of the connection identification function ahead of the time slot in which the first of the connection identification functions is arranged, a plurality of time slots on the scheduling table When a connection is reserved for overlapping the above-described connection identification function, the second or later connection identification function connected in a list with respect to one time slot on the scheduling table is adjacent to the forward or backward time. If the other connection identification function is already arranged in this time slot at the front or rear at this time, it is slid from the rear or the front to the end of the list of the connection identification functions. As a result of this The number of the connection identification feature disposed in the slot is limited to 0 or 1, thereby making it possible to perform the procedure for determining the cell read time slot.
[0045]
When reserving the first time slot of each connection and the ideal time slot arrangement starting from this first time slot, the time slot fluctuation allowable value of the connection bandwidth information corresponding to the connection identification function to be arranged is set. When setting and distributing the connection identification functions connected in a list form and arranging them in neighboring time slots, refer to the information on the allowable value of the connection indicated by the connection identification function included in the list, and A third step of rearranging the connection identification functions so that the values are in ascending order; and the second and subsequent ones of the rearranged list are slid to an adjacent rear time slot, and the slide of the connection identification function A fourth step that reduces the tolerance by the slide timing. Run the flop, when already the connection identification feature behind the time slot in this case is arranged, it is possible to again perform the steps to perform the third and fourth step.
[0046]
Instead of the part handling a single connection as a scheduling unit, a connection group composed of one or more connections can be handled as a scheduling unit.
[0047]
In the following, embodiments of the present invention will be described in more detail.
[0048]
(First Example)
The first embodiment of the present invention will be described below.
[0049]
First, for each connection, the number of cell arrangements within one period of the scheduling table and the ideal cell output interval are calculated according to the allocated bandwidth. If one cycle of the calculated time slot of each connection is called a time slot group of this connection, since the arrangement start position of the time slot group in the scheduling table is arbitrary, the entire distribution of time slots becomes uniform. Thus, the arrangement start position of the time slot group is determined.
[0050]
As time slot group arrangement start position determining means effective for uniformly arranging time slot positions, allocation is performed in order from the connection having the largest number of time slots included in the time slot group, and the time slots having the same number of time slots are included. For example, the group may be arranged by shifting the arrangement start position evenly.
[0051]
Note that only a single connection can be assigned to each time slot, but in the above method, a plurality of connections may be assigned to a single time slot. Therefore, by duplicating the time slot positions arranged at the ideal cell output interval within the arrangement position fluctuation allowable range, duplication of connection assignment to the time slots is avoided.
[0052]
FIG. 5 maps the cell output timing corresponding to the ideal cell output interval of each connection into an m × n array based on the method described above. Here, the number of connections is m, and the number of time slots for one period of the scheduling table is n. However, the cell output time slots of each connection are arranged at ideal calculated intervals, and there are portions where the same time slot is shared among a plurality of connections.
[0053]
Thus, in order to generate a scheduling table, an array (FIGS. 6 and 7) for constructing an arrangement list is prepared. Each of the arrays in FIGS. 6 and 7 is composed of elements of the number of time slots in which cells can be arranged. Hereinafter, the array in FIG. 6 is referred to as a time slot array, and the array in FIG. 7 is referred to as a connection array.
[0054]
The time slot arrangement corresponds to the time slot position of the scheduling table. This array has a pointer field, and the pointer of the element corresponding to the time slot position indicates the corresponding connection array element. Also, it is assumed that it has a ring-like structure logically, and continues to the next element after the last element.
[0055]
On the other hand, the connection array has a field for each connection ID and a field for pointers. It is pointed by the pointer of the time slot array. In addition, when a plurality of connections are assigned to the same time slot, a pointer is held to indicate the corresponding connection (element of connection array) in a list structure. In the initial state, initial values (for example, 0) are set for all elements of the time slot array and the connection array.
[0056]
According to the example of FIG. 5, the ideal time slot position of each connection is calculated. Since connection # 1 has time slot 1 as the ideal position, the pointer value of the first element of the time slot array is referred to. However, since an initial value (= 0) is set for this element, it can be seen that no connection has been placed yet. Therefore, the connection ID is written to an unused element of the connection array, and the element number of the array is written as a pointer to the first element of the time slot array. At this time, the top element of the time slot array is (1), and the top element of the connection array is (1: 0). Since connection # 1 also has time slot 5 in the ideal position, the fifth element of the time slot array is searched. Similarly, the element of time slot array is (2) and the second element of connection array is (1: Write 0).
[0057]
When the calculation of the time slot position of connection # 1 is completed, the next connection is calculated. Since the connection # 2 is also arranged in the time slot 1 (FIG. 5), first, the first element of the time slot array is referred to. However, since the pointer of the first element is not the initial value (= 0), it can be seen that another connection is already arranged. Therefore, the connection array indicated by this pointer is referred to. Here, since the pointer value of the first element of the time slot array is (1), the first element of the connection array is referred to.
[0058]
Since the first element of the connection array is (1: 0), connection # 1 has already been arranged, and since the pointer value is the initial value, no other connections can be arranged. Recognize. Therefore, the connection ID is written to an element of the vacant connection array, and the element number of the array is written as the pointer value of the connection array referred to earlier. That is, the first element is (1: 3).
[0059]
The contents of each array at this point are as shown in FIG. Thereafter, the same operation is performed for all connections. Conceptually, data having a list structure in which the terminal element pointer value is (0) as shown in FIG. 9 is constructed. The following description will be made using this list structure.
[0060]
When all the connections are arranged, connection IDs are arranged in the time slots based on the created time slot arrangement and connection arrangement.
[0061]
If only one connection ID is assigned to the corresponding time slot, it is arranged as it is, but if two or more connection IDs are assigned, the second and subsequent elements of the list connected by the pointer Reconnect the pointer to the next time slot array element. However, the moving connection array element is linked before the destination connection array element. This is to keep the maximum value of the deviation of the time slot position from the ideal arrangement position small. FIG. 10 shows this procedure.
[0062]
As described above, since the time slot array has a ring structure, when a plurality of connection IDs are assigned to the last array element, the rear connection array element is shifted to the first element of the time slot array. It will be. By this operation, the connection between the end portion and the head portion of the scheduling table can be performed seamlessly. FIG. 11 shows this procedure.
[0063]
In the above procedure, a plurality of connection array elements may be connected by a link structure. The maximum number of connections of the connection array elements matches the maximum deviation from the ideal arrangement position. Therefore, by monitoring the maximum value of this array element, it is possible to determine whether or not the permissible value of cell arrangement position fluctuation defined in advance is exceeded. For example, in the case of the arrangement of FIG. 12, if the maximum cell arrangement position fluctuation is 3 cells and the cell arrangement position fluctuation allowable value is 2 cells, the time slot of (18: 0) is the cell arrangement position fluctuation. The cell placement position is violated because the allowable value is exceeded.
[0064]
By repeating the above procedure, finally, all the connection array elements can be associated with the time scheduling elements on a one-to-one basis. The combination of the time slot position and the connection ID can be used as a scheduling table.
[0065]
As one of the criteria for determining the characteristics of the generated scheduling table, the maximum cell arrangement position fluctuation can be used. This value varies depending on the table allocation process (FIG. 5) of each connection performed first. Therefore, it is possible to generate a scheduling table with smaller cell arrangement position fluctuations by generating a plurality of patterns, performing the above-described scheduling table arrangement processing for each pattern, and comparing the maximum cell arrangement position fluctuations.
[0066]
(Second embodiment)
In the first embodiment of the present invention, the same maximum cell arrangement position fluctuation tolerance is set for all connections. However, when it is necessary to set a plurality of fluctuation tolerances, the arrangement position fluctuation is reduced. It needs to be managed more strictly.
[0067]
In this example, as shown in FIG. 13, the arrangement position fluctuation allowable value field is extended to the connection arrangement. In this field, the maximum deviation allowed for the ideal arrangement position calculated by calculation is set.
[0068]
The operation for mapping the ideal arrangement position calculated for each connection for each time slot array and each connection array does not depend on the above-described expansion, and thus the same operation as that in the first embodiment is performed. However, it is assumed that the arrangement position fluctuation allowable value set for each connection is set in addition to the connection ID in the set connection array.
[0069]
In the first embodiment, since the arrangement position fluctuation allowable value is the same for all connections, only the number of consecutive array elements is managed when a plurality of connection IDs are assigned to one time slot. It was. However, in the second embodiment, since the arrangement position fluctuation allowable value is different for each connection, the simple shift processing as in the first embodiment cannot be performed. Therefore, when a plurality of connection array elements are connected to the time slot array element, the rearrangement operation is performed in order from the smallest arrangement position fluctuation allowable value set in each connection array element (FIG. 14).
[0070]
At this time, it is necessary to monitor that the order of the connection array elements in the time slot array does not exceed the allowable arrangement position fluctuation value of each connection array element. If this condition cannot be satisfied, the finally generated time scheduling table becomes an illegal arrangement position fluctuation allowable value violation, and it is necessary to redo the mapping of the ideal arrangement position of each connection.
[0071]
By the above operation, the order other than the first connection array element in the time slot array is shifted to the next time slot array element. Here, the array position fluctuation allowable value of the connection array element to be shifted is decremented by the amount shifted (FIG. 14C). Then, when one or more connection arrays are assigned to the next time slot array element, the rearrangement operation is performed according to the arrangement position fluctuation allowable value of each connection array element as before.
[0072]
By repeating the above-described operation, it is possible to construct a scheduling table that satisfies the permissible fluctuation value of each connection even when different arrangement position fluctuation permissible values are set for each connection.
[0073]
(Third embodiment)
In the second embodiment, when the ideal arrangement position of each connection was assigned to the same time slot, the connection arrangement element that could not be arranged in the corresponding time slot was shifted to the rear time slot arrangement element. By extending the shift direction with respect to the front, it is possible to further reduce the arrangement position fluctuation. However, regarding the two adjacent time slots included in the same time slot group, when the previous time slot is shifted backward and the subsequent time slot is shifted forward, the interval with the preceding time slot is examined, and the corresponding It must be confirmed that the interval with the time slot arranged immediately before the connection is wider than the allowable interval. If the interval between time slots is narrower than the prescribed interval, the later time slots must be shifted backwards. In this way, when the forward shift and the backward shift are used together, the calculation for the time slot arrangement becomes more complicated than the case of the backward shift alone, so that the processing takes time, but the maximum cell arrangement position fluctuation Can be reduced.
[0074]
(Fourth embodiment)
In the embodiments so far, each element of the connection array is treated as an ID associated with a single connection. In addition, even when a plurality of connections are grouped and handled as a single connection group, this connection group can be replaced with a single connection in the above-described embodiment. That is, it is possible to assign an ID to each connection group and assign the bandwidth in proportion to the bandwidth set for each group.
[0075]
【The invention's effect】
As described above, according to the present invention, it is possible to generate a scheduling table having characteristics equivalent to or higher than those of the conventional scheduling method at a higher speed than the conventional method. It is possible to cope with the introduction of the time slot arrangement position fluctuation allowable value for each connection.
[Brief description of the drawings]
FIG. 1 is a block diagram of a cell read scheduling apparatus according to an embodiment of the present invention.
FIG. 2 is a flowchart showing a procedure of a cell read scheduling method according to an embodiment of the present invention.
FIG. 3 is a conceptual diagram of scheduling in one period of the scheduling table according to the embodiment of the present invention.
FIG. 4 is a flowchart showing a procedure when a scheduling table is created a plurality of times in order to create a scheduling table with little deviation from the ideal position according to the embodiment of the present invention.
FIG. 5 is a diagram showing a state in which cell output timings corresponding to ideal cell output intervals of each connection according to the embodiment of the present invention are mapped in an m × n array.
FIG. 6 is a diagram showing a time slot arrangement according to the embodiment of the present invention.
FIG. 7 is a diagram showing a connection arrangement according to an embodiment of the present invention.
FIG. 8 is a diagram showing a recording state of a time slot arrangement and a connection arrangement according to the embodiment of the present invention.
FIG. 9 is a diagram for explaining a scheduling method according to an embodiment of the present invention.
FIG. 10 is a diagram for explaining a scheduling method according to an embodiment of the present invention.
FIG. 11 is a diagram for explaining a scheduling method according to an embodiment of the present invention.
FIG. 12 is a diagram for explaining a scheduling method according to an embodiment of the present invention.
FIG. 13 is a diagram showing an expanded connection arrangement according to the embodiment of the present invention.
FIG. 14 is a diagram for explaining a scheduling method according to an embodiment of the present invention.
[Explanation of symbols]
1 Connection information holding means
2 Connection information sorting means
3 Connection identification means
4 List management means
5 Time slot management means
6 Initial allocation time slot reservation means
7 Ideal time slot reservation means
8 Arrangement time slot distribution means
9 Scheduling means
10 Cell reading means

Claims (3)

Connection information holding means for recording a set of identifier information and bandwidth information of each connection and fluctuation value information of ideal cell transmission timing of each connection ;
Connection information sorting means for rearranging a set of connection information recorded in the connection information holding means based on a designated policy;
A connection identifying means for identifying a connection allocated to each time slot arranged in each time slot on the scheduling table and in which the allowable value information for each connection is set ;
Means for connecting and managing connection identification means for connections to be placed in each time slot in a list for each time slot;
Time slot management means for accepting reservation of cell output timing required by each connection and arranging the connection identification means;
Reserving an ideal time slot arrangement of each connection starting from this leading time slot and the leading time slot of each connection to this timeslot managing means;
Comprising a scheduling means comprising means for causing dispersion in different time slots so when multiple connections for a single time slot on the scheduling table has reserved arranged overlapping the connection identification means,
Means for reading cells according to the scheduling result of the scheduling means ;
The connection information sorting means includes means for sorting in order from the wide band with reference to the bandwidth information of each connection,
The time slot management means includes means for accepting reservations for the ideal time slot arrangement of each connection starting from the first time slot and the first time slot from each connection according to the sorting order of the means for sorting,
The distributing means refers to the acceptance result of the accepting means, and when a plurality of connections overlap and reserve the connection identifying means for one time slot on the scheduling table, the allowable values are in ascending order. The first step of rearranging the connection identification means and the second and subsequent connection identification means of the rearranged list are slid to an adjacent rear time slot, and the allowable value of the slid connection identification means And the second step of decreasing the time by sliding timing, and when the other connection identifying means is already arranged in the rear time slot, the first and second steps are again performed. cell read, characterized in that it comprises means for performing the steps Scheduling apparatus. Refers to the bandwidth information of each connection, sorts in order from wide bandwidth, and accepts the reservation of the ideal time slot arrangement of each connection starting from the first time slot and the first time slot from each connection according to this sort order ,
Each time slot on the scheduling table is provided with a connection identification means for identifying the connection that reserved the time slot,
Connect and manage the connection identification means of the connection to be placed in each time slot in a list for each time slot,
When a plurality of connections reserve and arrange the connection identification means redundantly for one time slot on the scheduling table, among the connection identification means connected in a list for one time slot on the scheduling table The second and subsequent ones are slid into the adjacent rear time slot. At this time, when another connection identification means is already arranged in the rear time slot, the other connection identification means is moved from the front. Cell read scheduling which moves to the end of the list of the connection identification means slid and limits the number of connection identification means arranged in each time slot to 0 or 1 as a result, thereby determining a cell read time slot A method,
When reserving the first time slot of each connection and the ideal time slot arrangement starting from this first time slot, the time slot fluctuation allowable value of the corresponding connection bandwidth information for the connection identification means to be arranged is set. Set,
When distributing the connection identification means connected in a list form and arranging them in time slots within an allowable range of placement fluctuation , refer to the information on the allowable value of the connection indicated by the connection identification means included in the list, and The third step of rearranging the connection identification means so that the allowable value is in ascending order, and the second and subsequent ones of the rearranged list are slid to the adjacent rear time slot, and the slide of the connection identification means A fourth step of decreasing the allowable value by the slide timing, and when the connection identifying means is already arranged in the rear time slot at this time, the third and fourth steps are performed again. A cell readout scheduling method characterized by:   3. The cell read scheduling method according to claim 2, wherein instead of a part handling a single connection as a scheduling unit, a connection group constituted by one or more connections is handled as a scheduling unit.

Images