JP6189258B2 - Time slot allocation method for optical TDM ring network - Google Patents

Description

  The present invention provides an optical TDM (Time Division Multiplexing) that flexibly realizes transmission of a path in an arbitrary band by periodically allocating time slots by synchronous control of an optical switch toward realization of a power-saving all-optical network. The present invention relates to a multiplexing ring network, and clarifies a time slot allocation method for efficiently using the optical fiber bandwidth nearly 100% in an optical TDM ring network under arbitrary conditions.

A one-way optical TDM ring network is formed by connecting a plurality of nodes with optical fibers (hereinafter also referred to as links) in which the direction in which an optical signal is transmitted is determined in one direction. Each node has an optical switch (optical add / drop switch) inside it, which forwards information on the optical fiber to the next node, adds information to the optical fiber (add), and information from the optical fiber. The operation of dropping is performed periodically.
Unidirectional optical TDM ring network that sends optical signals in the forward direction and unidirectional optical TDM that sends optical signals in the reverse direction
By combining ring networks, a bidirectional optical TDM ring network capable of transmitting optical signals in both directions can be realized. For this reason, a technique for realizing a one-way optical TDM ring network is a basic technique of a general optical TDM ring network.

FIG. 3 shows an example of a unidirectional optical TDM ring network in which four nodes are connected by an optical fiber f. Each node is numbered from N0 to N3. Further, optical fibers f used for one-way communication from the node N0 to the node N1, from the node N1 to the node N2, from the node N2 to the node N3, and from the node N3 to the node N0 are respectively connected to the link L0-1 and the link L1-. 2, called link L2-3 and link L3-0.
Further, a delay adjusting optical fiber fd for adjusting the propagation delay time of one round of the ring to be an integral multiple of the time slot is inserted into the link L3-0. In the example of FIG. 3, it is assumed that the propagation delay time of one round of the ring is adjusted to 2 time slots.

  FIG. 4 shows an operation example of the optical switch (optical add / drop switch) SW in each node. FIG. 4A shows an operation of transferring the information α arriving from the previous node to the next node. FIG. 4B shows an operation of dropping (extracting) information β arriving from the previous node at the node and simultaneously adding (adding) information γ from the node to the next node.

Each information transfer is performed in units of a fixed length called a time slot. Information transfer in units of time slots is performed on each link. For this reason, the communication band required between nodes is implement | achieved by allocating only the required number of this time slot periodically with respect to communication between each node.
Specifically, when the line capacity of the link is C (bps) and the unit rate of path allocation is v (bps), M = (C / v) time slots are considered as units of periodic allocation. . Here, M is an integer, and M time slots, which are units of periodic time slot allocation, are called frames. Conversely, one frame has M time slots.

Assigning all the time slots in one frame period to communication between one node corresponds to assigning the entire line capacity C. Further, if only one time slot in one frame period is allocated to communication between one node, a bandwidth of 1 / M of the line capacity C is allocated.
For example, in a ring network that performs 10 Gbps transmission on a link, when performing bandwidth allocation in units of 1 Gbps for each path, M = 10/1 = 10. A value common to all nodes in the ring is used as the value of the frame period.

  Each node is provided with a timer that is synchronized with each other with high accuracy, and an offset corresponding to a signal propagation delay from the node N0 is added to the timer of each node. Thus, an operation can be performed in which the start of the frame of each node starts just in time with the time when the optical signal transmitted in the start time slot of the frame of the node N0 passes through each node.

FIG. 3 also shows a specific example of periodic time slot allocation on each link when M = 4. Numbers # 0 to # 3 are repeatedly assigned to the time slots on each link as positions in the frame. In each time slot, a path assigned to use the time slot is described. A blank timeslot has not yet been assigned any path.
Here, since the periodic time slot assignment is performed, the same communication is assigned to the time slot of the same number on each link.

  As a specific example, considering the case of assigning a path from the node N0 to the node N3 (in FIG. 3, this path is indicated by “0 → 3”), this path is represented by a link L0-1 and a link L1-2. , Via link L2-3. Therefore, in the example of FIG. 3, the time slots # 0 and time slots # 2 of these links L0-1, L1-2, and L2-3 are allocated to the path from the node N0 to the node N3.

Similarly, for communication from the node N3 to the node N2, the time slot # 3 of the link L3-0 and the time slot # 1 of the link L0-1 and the link L1-2 are allocated.
Here, the time slot numbers are changed between the link L3-0 and the link L0-1, because the propagation delay time of one round of the ring is two time slots, so the time slot on the link L3-0 is This is because on the link L0-1, the time slot number is transferred to the time slot which is two larger.
However, since the time slot number is a frame period, the time slot number shifted by 2 time slots from the time slot of # 3 moves to # 1, which is a remainder obtained by a division operation of (3 + 2) ÷ 4. become.

  Similarly, time slot # 1 of link L2-3, link L3-0, and time slot # 3 of link L0-1 are assigned to communication from node N2 to node N1.

In this way, when a communication request between arbitrary nodes is given, assigning time slots within a frame period as efficiently as possible while considering that there is no collision with other communication is possible. This is an important issue in the operation of the TDM ring network.
As will be described later, the present invention is a method for performing very efficient time slot allocation in such time slot allocation.

Non-patent document 1, which is a prior art document, shows the following. That is, in the one-way optical TDM ring in which the propagation delay time of one round of the ring is D, when assigning time slots with a frame period F = M (where M ≧ D), the following (a) and (B) is shown to be equivalent.
(A) Assigning a time slot to any communication between nodes.
(B) Time slot allocation is performed in an area of continuous K time slots called “time slot cycles”, which has a length of M / K circumferences with respect to the length of one ring circumference. about.
Here, K is the greatest common divisor of D and M. As described above, M is an integer, and the propagation delay time D is defined by the number of time slots. This time slot cycle is a logical configuration resulting from performing periodic time slot assignments.

Non-patent document 2, which is a prior art document, shows an extension of the idea shown in non-patent document 1 to a bidirectional optical TDM ring network.
Note that the names of Non-Patent Literature 1 and Non-Patent Literature 2 are described later.

Below, the outline | summary of the time slot allocation method shown by the nonpatent literature 1 is shown.
5 (a) to 5 (c), the time slots of each link shown in FIG. 3 are arranged vertically, and each time slot that transmits information and the transmitted information pass next. The time slot is connected with an arrow.

  FIG. 5A shows a state when the propagation delay time D for one round of the ring is 2 time slots, and FIG. 5B shows a state when the propagation delay time D for one round of the ring is 3 time slots. FIG. 5 (c) shows a state when the propagation delay time D for one round of the ring is 4 time slots.

First, the state of FIG.
As described above, since the frame start timing in each link is synchronized, the information of each time slot is the next link from link L0-1 to link L1-2, link L2-3, and link L3-0. Pass the same numbered time slot above.

However, since the propagation delay time D for one round of the ring is two time slots, the information transmitted in the time slot #i (i = 0, 1, 2, 3) of the link L3-0 Time slot # ((i + 2) mod 4) will be passed. Specifically, the time slot # 0 shifts to # 2, # 1 shifts to # 3, # 2 shifts to # 0, and # 3 shifts to # 1.
As a result, referring to FIG. 5A, a time slot cycle formed by connecting the time slot # 0 column and the # 2 column to each other, and the time slot # 1 column and the time slot # 3 column are formed. It can be seen that two time slot cycles are formed, which are time slot cycles formed by being connected to each other.

  Note that “mod” means “divisor”. Therefore, ((i + 2) mod 4) means “remainder” = “remainder” obtained by the division operation of (i + 2) ÷ 4.

As another example, considering the case where the propagation delay time D of the ring 1 round is 3 time slots, the time slot on the link L3-0 is changed to the time slot shifted by 3 time slots on the link L0-1. Because of the connection, the continuous relationship of similar time slots is shown in FIG. 5B.
In this case, the sequence of time slot # 0 is connected to the sequence of time slot # 3, and then returns to time slot # 0 via time slot # 2 and time slot # 1, thereby forming one continuous time slot cycle. You can see that

  As yet another example, considering a case where the propagation delay time D of the ring 1 round is 4 time slots, the time slot on each link L3-0 is shifted by 4 time slots on the link L0-1. Since it is connected to the slot, it returns to the same time slot, as shown in FIG. 5C, resulting in four independent time slot cycles.

  In general, in an optical TDM ring that performs periodic time slot assignment, information sent from time slot #i (i = 0, 1,... M−1) of node N0 is dropped at any node. If the transfer is continued without any change, the next pass passes through the time slot # ((i + D) mod M) of the node N0. Further, after k laps (k ≧ 1), it passes through the time slot # ((i + k · D) mod M) of the node N0.

  Therefore, considering the condition when this information first returns to the original time slot #i, (k · D mod M) = 0 from ((i + k · D) mod M) = i. This means that k · D is the least common multiple of D and M. Therefore, when K is the greatest common divisor of D and M, it can be seen that the above condition is satisfied when k = M / K.

Accordingly, the number of times k of the time slot cycle as shown in FIG. 5 is (M / K). At this time, it is known that the number of time slot cycles is K.
In fact, in each of FIGS. 5A, 5B, and 5C, the greatest common divisors of M and D are K = 2, 1, and 4, respectively, and the number of time slot cycles is K. It can be confirmed that the number of times k of the time slot cycle is 4/2, 4/1, and 4/4, respectively.

This state is shown in the table below.

  In Non-Patent Document 1, after understanding the structure of a logical time slot cycle associated with such periodic time slot allocation, a time slot when a bandwidth request for a path for transmission between arbitrary nodes is given. The assignment method is shown. FIG. 6 shows a conventional time slot allocation procedure.

First, a propagation delay time D (a time defined by the number of time slots) for one round of the ring of the ring network, a division number M for dividing the line capacity of the link when assigning time slots to paths,
A requested bandwidth matrix T whose elements are the number of bandwidth time slots (number of requested time slots) Tij for the path from each node Ni to the node Nj is input (steps 1, 2, and 3 in FIG. 6).

  Next, using the calculation method described above, the frame period F = M, the frame period F, the number k of times of the time slot cycle when the propagation delay time D is set (= (M / K)), and the time slot cycle Is derived (= the greatest common divisor of M and D) (steps 4 and 5 in FIG. 6).

  Next, based on the derived time slot cycle, a time slot is assigned to each path in the following procedure (step 6 in FIG. 6).

(Procedure 1)
All time slots are left unassigned. Assume that the number of requested time slots from the node Ni to the node Nj is given by Tij. Here, one arbitrary path is selected from the unassigned paths, and this path is defined as a path (i, j) from the node Ni to the node Nj.

(Procedure 2)
One unassigned time slot cycle is selected, and this is called an assigned time slot cycle y. If there are no unassigned time slot cycles, the procedure ends.
Allocate a time slot for path (i, j) on time slot cycle y. Here, for an unassigned time slot cycle, any one path can be assigned without fail.
Further, after assigning one time slot of the time slot cycle y to one path (i, j), Tij = Tij−1. That is, the number of requested time slots is reduced by one.

(Procedure 3)
One path starting from a node as close as possible to the node Nj is selected from the unassigned paths. Specifically, the path starting from the node Nj is first searched from among the unassigned paths. If the path is not found, the path starting from the node Nj + 1 is detected. Search for unassigned paths while shifting. If a path is found, this is newly set as a path (i, j). If no unassigned path is found, the procedure is terminated.

(Procedure 4)
The path (i, j) found in step 3 is traced in order from the time slot next to the time slot assigned immediately before in the currently assigned time slot cycle y, and if a free time slot is found, it is found in step 3. Assign the correct path. Then, after setting Tij = Tij-1, the process proceeds to step 3. If an empty time slot is not found, the procedure goes to step 2.

  When the above procedure is completed, if there are no unassigned paths, time slots are assigned to all paths. If an unassigned path remains, it becomes a path that could not be assigned in the above procedure.

  With the above-described procedure, the procedure shown in FIG. 6 is executed, and the time slot assignment is completed.

  As a specific example, consider a path request in which the number of requested time slots Tij between nodes is T03 = 2, T10 = 1, T21 = 1, T32 = 1, and all others are 0, and the above procedure is used to determine FIG. FIGS. 7A, 7B, and 7C show the results of path assignment based on the time slot cycles shown in a), (b), and (c). Here, in any case, the path (0, 3) is selected first, and the assignment from the time slot # 0 of the link L0-1 is shown.

  For example, T03 = 2 indicates that the number of time slots necessary for the path transmitted from the node N0 to the node N3 (the number of requested time slots) is two. For example, the path (0, 3) represents a path transmitted from the node N0 to the node N3. In FIG. 7, this path is indicated by “0 → 3”.

  For example, in FIG. 7A, the path (0, 3) is first assigned from the time slot # 0 of the link L0-1, and then the time slot to the path (3, 2) is assigned to the time slot # 0 of the link L3-0. Assigned from.

  Next, when the path (2, 1) is assigned, the time slot cycle consisting of the time slots # 0 and # 2 cannot secure the time slots continued on the links L2-3, L3-0, and L0-1. A path (2, 1) is assigned to the second time slot cycle consisting of time slot # 1 and time slot # 3.

  Subsequently, after assigning the path (1, 0) from the time slot # 3 of the link L1-2, the procedure is completed without continuously assigning the path (0, 3) in the remaining empty time slots. It becomes.

  On the other hand, when D = 3, requests for all paths can be continuously assigned to one time slot cycle (see FIG. 7B).

  In addition, when D = 4, each time slot cycle is a short time slot cycle of one round, so only one path can be accommodated in each time slot cycle (see FIG. 7C). . For this reason, the procedure is completed without assigning the second time slot of the path (0, 3).

As seen in the example above, if all available time slots are one, such as when the time slot cycle is D = 3, the time slots to each path are consecutive. Can be allocated without waste (see FIG. 7B).
However, when the time slot cycle is divided into a plurality of times as in the case of D = 2 or D = 4, even if several time slots are distributed and vacated in different time slot cycles, this is a continuous time. Since it cannot be assigned to a path as a slot, it can be seen that the time slot cannot be used effectively (see FIGS. 7A and 7C).

A. Chen, C.−T.Lea, and AK−S.Wong, “A New Optical TDM Ring Architecture,” IEEE Trans. Commun., Vol. 55, no. 11, pp. 2134-2141, Nov. 2007 . A. Chen, AK−S.Wong, and C.−T.Lea, “On the Space Reuse Efficiency of TSI−free OTDM Rings,” IEEE Commun., Lett., Vol. 12, no. 4, pp. 322 −324, Apr., 2008.

As described above, in an optical TDM ring network that performs periodic time slot assignment, the propagation delay time D (time defined by the number of time slots) of one round of the ring and the line capacity when assigning time slots to paths are as follows. Depending on the number of divisions M, the number of laps and the number of timeslot cycles are determined, and depending on the number of timeslot cycles, there is a problem that efficient allocation may not be performed for a bandwidth request of an arbitrary path. It has become.
The propagation delay time D is determined from physical conditions such as the length of the optical fiber used by the ring network, and the division number M of the line capacity when assigning the time slot to the path indicates how the line capacity of the ring network is determined. Since it is determined by the operational policy of whether to use, it is difficult to set each to a free value.

In such a situation, if a given combination of D and M values is bad and has a large greatest common divisor, the ring network can always perform only inefficient time slot allocation.
As described above, for example, when D = M, the time slot cycle is a short cycle (see FIGS. 5 (c) and 7 (c)), so it is assigned to one time slot cycle. Only a very inefficient time slot allocation can be achieved with a small number of paths.

  In view of the above prior art, the present invention always performs efficient time slot allocation even if the number of divisions M of the line capacity in the optical TDM ring network and the propagation delay time D around one ring are arbitrary values. It is an object of the present invention to provide a time slot allocation method for an optical TDM ring network.

The present invention for solving the above problems
A plurality of nodes having optical add / drop switches are connected by a plurality of links made of optical fibers. The start timings of the frames in the plurality of links are synchronized, and the propagation delay time D of one round of the ring is In an optical TDM ring network that is an integral multiple of time slots, an optical TDM ring network that periodically allocates the required number of time slots to the required number of time slots required for a path transmitted between arbitrary nodes. A time slot allocation method,
Inputting the propagation delay time D, a division number M for dividing the line capacity of the link when assigning time slots to paths, and a requested bandwidth matrix T having the number of requested time slots as an element;
Inputting an arbitrary natural number as the period multiple P;
Obtaining the frame period F by calculating F = M × D × P;
Based on the obtained frame period F and the propagation delay time D, the number of time slot cycles K = D, which is the greatest common divisor of F and D, and the number of times of the time slot cycle obtained by dividing F by K = M A step of obtaining xP;
The number of paths equal to the number of requested time slots of the requested bandwidth matrix T is obtained by multiplying by P, and K = D timeslots where the number of laps is k = M × P with respect to the traffic path. Assigning cycle time slots;
It is characterized by having.

The present invention also provides
In the time slot allocation method,
The step of allocating K = D time slots of the time slot cycle, where the number of laps is k = M × P, is assigned to the traffic path,
A first procedure of selecting an arbitrary path from among the unassigned paths and setting this path as a path (i, j) from the node Ni to the node Nj;
One unassigned time slot cycle is selected, which is set as the time slot cycle y being assigned, and a time slot for the path (i, j) is assigned on the time slot cycle y, and after this assignment, the node Ni to the node Nj A second procedure for reducing the number of requested time slots Tij by 1 and terminating the procedure if there are no unassigned time slot cycles;
A third procedure that selects one path starting from a node as close as possible to the node Nj from among the unassigned paths, sets the selected path as a new path (i, j), and ends the procedure if no unassigned path is found. When,
In the time slot cycle y being allocated, the time slot is sequentially traced from the time slot next to the time slot allocated immediately before, and if an empty time slot is found, the new path (i, i, j), and after this allocation, the requested time slot number Tij is decremented by 1, and then the process proceeds to the third procedure, and if no empty time slot is found, the fourth procedure proceeds to the second procedure;
It is characterized by having.

The present invention also provides
A plurality of nodes having optical add / drop switches are connected by a plurality of links made of optical fibers. The start timings of the frames in the plurality of links are synchronized, and the propagation delay time D of one round of the ring is In an optical TDM ring network that is an integral multiple of time slots, an optical TDM ring network that periodically allocates the required number of time slots to the required number of time slots required for a path transmitted between arbitrary nodes. A time slot allocation method,
Inputting the propagation delay time D, a division number M for dividing the line capacity of the link when assigning time slots to paths, and a requested bandwidth matrix T having the number of requested time slots as an element;
Inputting an arbitrary natural number as the period multiple P;
Obtaining the frame period F by calculating F = M × D × P + 1;
Based on the determined frame period F and the propagation delay time D, the number of time slot cycles K = 1, which is the greatest common divisor of F and D, and the number of times of the time slot cycle obtained by dividing F by K = M A step of obtaining × D × P + 1;
The number of times of traffic obtained by multiplying the number of requested time slots of the requested bandwidth matrix T by P × D is obtained, and the number of laps k = M × D × P + 1 for the traffic path is K = 1. Assigning a time slot of said time slot cycle;
It is characterized by having.

The present invention also provides
In the time slot allocation method,
The step of assigning a time slot of K = 1 time slot cycle, where the number of laps is k = M × D × P + 1, to the traffic path,
A first procedure of selecting an arbitrary path from among the unassigned paths and setting this path as a path (i, j) from the node Ni to the node Nj;
One time slot cycle is assigned as a time slot cycle y, and a time slot for the path (i, j) is assigned on the time slot cycle y. After this assignment, the requested number of time slots Tij from the node Ni to the node Nj A second step of subtracting 1 from
A third procedure that selects one path starting from a node as close as possible to the node Nj from among the unassigned paths, sets the selected path as a new path (i, j), and ends the procedure if no unassigned path is found. When,
In the time slot cycle y being allocated, the time slot is sequentially traced from the time slot next to the time slot allocated immediately before, and if an empty time slot is found, the new path (i, i, j), and after this assignment, the requested time slot number Tij is decremented by 1, and then the fourth procedure proceeds to the third procedure;
It is characterized by having.

In the time slot allocation method of the optical TDM ring network of the present invention, even when the number of divisions M of the line capacity and the propagation delay time D of one round of the ring in the optical TDM ring network are arbitrary values, an efficient time slot is always obtained. Assignment can be possible.
Therefore, even if the values of M and D are determined by the environment of the ring network to be used, the operation policy of the ring network, etc., and the values of M and D cannot be adjusted to any convenient value. By using the present invention, a highly efficient ring network can be used.

2 is a flowchart illustrating a time slot allocation method for an optical TDM ring network according to the first embodiment of the present invention. 9 is a flowchart illustrating a time slot allocation method for an optical TDM ring network according to a second embodiment of the present invention. The block diagram which shows the structure of a one-way optical TDM ring network, and the state of the time slot allocation in each link. The block diagram which shows the operation example of an optical add / drop switch. Explanatory drawing which shows the formation state of a time slot cycle. 9 is a flowchart showing a time slot allocation method for an optical TDM ring network according to the prior art. Explanatory drawing which shows the state which allocates a time slot based on a time slot cycle.

Embodiments of the present invention will be described below.
As can be seen from the example shown in FIG. 7 described above, when the time slot cycle is long (when the number of times k of the time slot cycle is large), efficient time slot allocation becomes possible (FIG. 7B). )), And a condition for lengthening the time slot cycle when the division number M of the line capacity and the transmission delay time D of one round of the ring are given.
Specifically, the value of the frame period F is not the conventional M, but a value that is an integral multiple of M × D or a value obtained by adding 1 to an integral multiple of M × D.
In other words, the frame period F is conventionally set to M = M, but in the present embodiment, the frame period F = M × D × integer or the frame period F = M × D × integer + 1.

  First, when the period multiple P is an arbitrary natural number (a natural number excluding 0), a time slot cycle when the frame period F is not M but M × D × P is obtained.

  At this time, since the frame period F is an integral multiple of D regardless of M or P, the greatest common divisor of the frame period F and the ring propagation delay time D is D. Accordingly, at this time, the number of times k of the time slot cycle is M × P, and the number K of time slot cycles is D.

In summary, the greatest common divisor between the frame period F (= M × D × P) and the ring propagation delay time D is the number K of time slot cycles. Therefore, as described above, in this embodiment, the number of time slot cycles K = D. Incidentally, conventionally, K is the greatest common divisor of M and D.
Further, the number of times k of the time slot cycle is obtained by dividing the frame period F (= M × D × P) by the number of time slot cycles K (= D). As described above, in the present embodiment, the number of times of the time slot cycle is k = MP. Incidentally, k = M / K in the prior art.

  Here, when the required number of time slots required for the path from the node Ni to the node Nj is given by the required bandwidth matrix T = {Tij}, consider the traffic obtained by multiplying each of these elements by P, and this is expressed as M × P described above. Allocate for a perimeter length time slot cycle. The remaining D-1 other time slot cycles are allocated in the same manner as this allocation.

  As a result, for one path, P time slots are allocated to each time slot cycle, and D timeslots are allocated. Therefore, in the frame period F = M × D × P, , P × D time slots are allocated.

  Therefore, time slots can be assigned at the same ratio as when one time slot is assigned to the original frame period F = M. It is thought that there is. Also, the time slot cycle length (number of laps) is M * P, which is longer than M / K in the past, so it is possible to efficiently allocate tracks to paths. is there.

On the other hand, considering the case where the frame period F is M × D × P + 1 instead of M, the greatest common divisor of F and D is 1. In this case, the number of time slot cycles K is 1, and the number of times k of the time slot cycle is M × D × P + 1. Therefore, a very long time slot cycle can be realized regardless of M or D.

In summary, the greatest common divisor between the frame period F (= M × D × P + 1) and the ring propagation delay time D is the number K of time slot cycles. Therefore, as described above, in this embodiment, the number of time slot cycles K = 1. Incidentally, conventionally, K is the greatest common divisor of M and D.
The number of times k of the time slot cycle is obtained by dividing the frame period F (= M × D × P + 1) by the number of time slot cycles K (= 1). As described above, in this embodiment, the number of times of the time slot cycle is k = M × D × P + 1. Incidentally, k = M / K in the prior art.

  Here, the traffic to be allocated is calculated by multiplying the original traffic by D × P, and this is allocated to a long time slot cycle of M × D × P + 1 rounds. In such time slot allocation, since the time slot cycle is very long, efficient time slot allocation is possible.

The state of the above two embodiments is shown in the table as follows.

Hereinafter, a method for assigning time slots in an optical TDM ring network according to the present invention will be described in detail based on embodiments.
In the following embodiment, the propagation delay time D for one round of the ring, the division number M of the line capacity, the required bandwidth matrix T, and the period multiple P are input to the control device, and the respective calculation processes are performed. Time slot allocation is performed. However, since hardware such as a control device is a normal technique, description of the hardware is omitted, and only a data processing procedure is described.

[Example 1]
FIG. 1 shows a time slot allocation method according to Embodiment 1 of the present invention.
Example 1 shown in FIG. 1 is different from the conventional procedure shown in FIG.
(1) A procedure (step 14 in FIG. 1) for inputting an arbitrary natural number (natural number excluding 0) as the period multiple P;
(2) A procedure (step 15 in FIG. 1) for calculating the frame period F = M × D × P from the ring propagation delay D, the line capacity division number M, and the period multiple P;
(3) A procedure (step 17 in FIG. 1) for multiplying the elements of the requested bandwidth matrix T (the number of requested time slots Tij) by P is added.

  According to such a procedure, the above-described control is performed with the frame period F set to M × D × P, and the allocation of the required bandwidth to each path is performed with the D number of laps k being M × P time slots. It will be done using the cycle.

Here, an M × P time slot cycle that is enlarged P times the number of originally requested time slots is assigned.
In addition, since allocation is performed in the same manner for each of the D time slot cycles, it is possible to allocate P × D times as much traffic as the total frame period M × D × P.

  Next, the time slot allocation method according to the first embodiment will be described in further detail.

First, a propagation delay time D (a time defined by the number of time slots) for one round of the ring of the ring network, a division number M for dividing the line capacity of the link when assigning time slots to paths,
The requested bandwidth matrix T having the number of bandwidth time slots (number of requested time slots) Tij for the path from each node Ni to the node Nj is input (steps 11, 12, and 13 in FIG. 1).

Next, an arbitrary natural number is input as the period multiple P (step 14 in FIG. 1).
Subsequently, the calculation of the frame period F = M × D × P is performed (step 15 in FIG. 1).

  Based on the frame period F (= M × D × P) and the propagation delay time D, D time slot cycles in which the number of rounds k of the time slot cycle is M × P are calculated (step 16 in FIG. 1). ).

  The number of requested time slots (elements) Tij required for the path from the node Ni to the node Nj given to the requested bandwidth matrix T is expanded by P times (step 17 in FIG. 1).

  In this way, the traffic obtained by multiplying each element by P is considered, and this is assigned to the time slot cycle having the length of M × P. Then, the same assignment as this assignment is performed for the remaining D-1 other time slot cycles (step 18 in FIG. 1).

  Step 18 of FIG. 1 is performed according to the following procedure 11 to procedure 13.

(Procedure 11)
All time slots are left unassigned. Assume that the number of requested time slots from the node Ni to the node Nj is given by Tij. Here, one arbitrary path is selected from the unassigned paths, and this path is defined as a path (i, j) from the node Ni to the node Nj.

(Procedure 12)
One unassigned time slot cycle is selected, and this is called an assigned time slot cycle y. If there are no unassigned time slot cycles, the procedure ends.
Allocate a time slot for path (i, j) on time slot cycle y. Here, for an unassigned time slot cycle, any one path can be assigned without fail.
Further, after assigning one time slot of the time slot cycle y to one path (i, j), Tij = Tij−1. That is, the number of requested time slots is reduced by one.

(Procedure 13)
One path starting from a node as close as possible to the node Nj is selected from the unassigned paths. Specifically, the path starting from the node Nj is first searched from among the unassigned paths. If the path is not found, the path starting from the node Nj + 1 is detected. Search for unassigned paths while shifting. If a path is found, this is newly set as a path (i, j). If no unassigned path is found, the procedure is terminated.

(Procedure 14)
The path (i, j) found in step 13 is traced sequentially from the time slot next to the time slot assigned immediately before in the currently assigned time slot cycle y, and if a free time slot is found, the path is found in step 13. Assign the correct path. Then, after setting Tij = Tij-1, the process proceeds to step 13. If no empty time slot is found, the process proceeds to step 12.

  When the above procedure is completed, if there are no unassigned paths, time slots are assigned to all paths.

  According to the above procedure, the procedure shown in FIG. 1 is executed, and the time slot assignment is completed.

[Example 2]
FIG. 2 shows a time slot allocation method according to the second embodiment of the present invention.
The second embodiment shown in FIG. 2 is different from the conventional procedure shown in FIG.
(1) A procedure (step 24 in FIG. 2) for inputting an arbitrary natural number (natural number excluding 0) as the period multiple P;
(2) A procedure (step 25 in FIG. 2) for calculating the frame period F = M × D × P + 1 from the ring propagation delay D, the line capacity division number M and the period multiple P.
(3) A procedure (step 27 in FIG. 2) for multiplying the elements of the requested bandwidth matrix T (the number of requested time slots Tij) by P × D is added.

  According to such a procedure, the above-described control is performed with the frame period F set to M × D × P + 1, and the allocation of the required bandwidth to each path is performed such that one lap number k is M × D × P + 1. This is done using the time slot cycle.

  Here, an M × D × P + 1 time slot cycle that is enlarged by P × D times the number of originally requested time slots is assigned.

  Allocation of one time slot in the conventional M cycle is equivalent to allocating 1 / M bandwidth of the line capacity, but in the allocation in the second embodiment, D × P times in M × D × P + 1 cycle. Since slots are allocated, the line capacity in this case is (P × D) / (M × D × P + 1), which is a value slightly smaller than 1 / M. However, since a time slot is assigned to a path using a very large continuous time slot cycle, it is considered to be highly effective as a very efficient method.

  Next, the time slot allocation method according to the second embodiment will be described in further detail.

First, a propagation delay time D (a time defined by the number of time slots) for one round of the ring of the ring network, a division number M for dividing the line capacity of the link when assigning time slots to paths,
Also, a requested bandwidth matrix T whose elements are the number of bandwidth time slots (number of requested time slots) Tij for the path from each node Ni to the node Nj is input (steps 21, 22, and 23 in FIG. 2).

Next, an arbitrary natural number is input as the period multiple P (step 24 in FIG. 2).
Subsequently, the calculation of the frame period F = M × D × P + 1 is performed (step 25 in FIG. 2).

  Based on the frame period F (= M × D × P + 1) and the propagation delay time D, one time slot cycle in which the number of rounds k of the time slot cycle is M × D × P + 1 is calculated (FIG. 2). Step 26).

  The requested time slot number (element) Tij necessary for the path from the node Ni to the node Nj given to the requested bandwidth matrix T is expanded by P × D (step 27 in FIG. 2).

  Thus, consider the traffic obtained by multiplying each element by P × D, and assign it to one time slot cycle having a length of M × D × P + 1. (Step 28 in FIG. 2).

  Step 28 of FIG. 2 is performed according to the following procedure 21 to procedure 23.

(Procedure 21)
All time slots are left unassigned. Assume that the number of requested time slots from the node Ni to the node Nj is given by Tij. Here, one arbitrary path is selected from the unassigned paths, and this path is defined as a path (i, j) from the node Ni to the node Nj.

(Procedure 22)
One time slot cycle is referred to as an assigned time slot cycle y.
A time slot for the path (i, j) is allocated on this time slot cycle y.
Further, after assigning one time slot of the time slot cycle y to one path (i, j), Tij = Tij−1. That is, the number of requested time slots is reduced by one.

(Procedure 23)
One path starting from a node as close as possible to the node Nj is selected from the unassigned paths. Specifically, the path starting from the node Nj is first searched from among the unassigned paths. If the path is not found, the path starting from the node Nj + 1 is detected. Search for unassigned paths while shifting. If a path is found, this is newly set as a path (i, j). If no unassigned path is found, the procedure is terminated.

(Procedure 24)
The path (i, j) found in the procedure 23 is sequentially traced from the time slot next to the time slot assigned immediately before in the time slot cycle y. If an empty time slot is found, the path found in the procedure 23 is assigned. Then, after setting Tij = Tij−1, the process proceeds to step 23.

  When the above procedure is completed, if there are no unassigned paths, time slots are assigned to all paths.

  According to the above-described procedure, the procedure shown in FIG. 2 is executed, and time slot assignment is completed.

  The present invention can be used when allocating time slots in a one-way optical TMD ring network or a bidirectional optical TMD ring network.

N0, N1, N2, N3 Node L0, L1, L2, L3 Link f Optical fiber fd Delay adjusting optical fiber SW Optical add / drop switch

Claims (4)

A plurality of nodes having optical add / drop switches are connected by a plurality of links made of optical fibers. The start timings of the frames in the plurality of links are synchronized, and the propagation delay time D of one round of the ring is In an optical TDM ring network that is an integral multiple of time slots, an optical TDM ring network that periodically allocates the required number of time slots to the required number of time slots required for a path transmitted between arbitrary nodes. A time slot allocation method,
Inputting the propagation delay time D, a division number M for dividing the line capacity of the link when assigning time slots to paths, and a requested bandwidth matrix T having the number of requested time slots as an element;
Inputting an arbitrary natural number as the period multiple P;
Obtaining the frame period F by calculating F = M × D × P;
Based on the obtained frame period F and the propagation delay time D, the number of time slot cycles K = D, which is the greatest common divisor of F and D, and the number of times of the time slot cycle obtained by dividing F by K = M A step of obtaining xP;
The number of paths equal to the number of requested time slots of the requested bandwidth matrix T is obtained by multiplying by P, and K = D timeslots where the number of laps is k = M × P with respect to the traffic path. Assigning cycle time slots;
A time slot allocation method for an optical TDM ring network. In claim 1,
The step of allocating K = D time slots of the time slot cycle, where the number of laps is k = M × P, is assigned to the traffic path,
A first procedure of selecting an arbitrary path from among the unassigned paths and setting this path as a path (i, j) from the node Ni to the node Nj;
One unassigned time slot cycle is selected, which is set as the time slot cycle y being assigned, and a time slot for the path (i, j) is assigned on the time slot cycle y, and after this assignment, the node Ni to the node Nj A second procedure for reducing the number of requested time slots Tij by 1 and terminating the procedure if there are no unassigned time slot cycles;
A third procedure that selects one path starting from a node as close as possible to the node Nj from among the unassigned paths, sets the selected path as a new path (i, j), and ends the procedure if no unassigned path is found. When,
In the time slot cycle y being allocated, the time slot is sequentially traced from the time slot next to the time slot allocated immediately before, and if an empty time slot is found, the new path (i, i, j), and after this allocation, the requested time slot number Tij is decremented by 1, and then the process proceeds to the third procedure, and if no empty time slot is found, the fourth procedure proceeds to the second procedure;
A time slot allocation method for an optical TDM ring network. A plurality of nodes having optical add / drop switches are connected by a plurality of links made of optical fibers. The start timings of the frames in the plurality of links are synchronized, and the propagation delay time D of one round of the ring is In an optical TDM ring network that is an integral multiple of time slots, an optical TDM ring network that periodically allocates the required number of time slots to the required number of time slots required for a path transmitted between arbitrary nodes. A time slot allocation method,
Inputting the propagation delay time D, a division number M for dividing the line capacity of the link when assigning time slots to paths, and a requested bandwidth matrix T having the number of requested time slots as an element;
Inputting an arbitrary natural number as the period multiple P;
Obtaining the frame period F by calculating F = M × D × P + 1;
Based on the determined frame period F and the propagation delay time D, the number of time slot cycles K = 1, which is the greatest common divisor of F and D, and the number of times of the time slot cycle obtained by dividing F by K = M A step of obtaining × D × P + 1;
The number of times of traffic obtained by multiplying the number of requested time slots of the requested bandwidth matrix T by P × D is obtained, and the number of laps k = M × D × P + 1 for the traffic path is K = 1. Assigning a time slot of said time slot cycle;
A time slot allocation method for an optical TDM ring network. In claim 3,
The step of assigning a time slot of K = 1 time slot cycle, where the number of laps is k = M × D × P + 1, to the traffic path,
A first procedure of selecting an arbitrary path from among the unassigned paths and setting this path as a path (i, j) from the node Ni to the node Nj;
One time slot cycle is assigned as a time slot cycle y, and a time slot for the path (i, j) is assigned on the time slot cycle y. After this assignment, the requested number of time slots Tij from the node Ni to the node Nj A second step of subtracting 1 from
A third procedure that selects one path starting from a node as close as possible to the node Nj from among the unassigned paths, sets the selected path as a new path (i, j), and ends the procedure if no unassigned path is found. When,
In the time slot cycle y being allocated, the time slot is sequentially traced from the time slot next to the time slot allocated immediately before, and if an empty time slot is found, the new path (i, i, j), and after this assignment, the requested time slot number Tij is decremented by 1, and then the fourth procedure proceeds to the third procedure;
A time slot allocation method for an optical TDM ring network.

Images