JP2013236124A - Packet scheduling method and communication device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a packet scheduling method that can change a quantum value without changing a ratio among flows according to the state of loads of other processing.SOLUTION: A packet scheduling method for determining a packet to be transmitted at each schedule round by use of an EBRR scheme is able to change a quantum value without changing a ratio among respective flows in concert with the state of loads of other processing performed by a CPU and the like when scheduling processing is implemented by means of software by the CPU and the like.

Description

  The present invention relates to a packet scheduling method and a communication apparatus.

  As a packet scheduling algorithm, a plurality of O (1) scheduling schemes with a calculation amount of O (1) have been proposed. Examples include Deficit Round Robin (DRR), Surplus Round Robin (SRR), Eligibility Based Round Robin (EBRR) (see Non-Patent Document 1 below), and the like.

  Among these, EBRR proposed by Lenzini et al. Differs from DRR and SRR in that the amount of calculation is O (1) even when the quantum value (value added to the Credit Counter of each flow for each scheduling round) is less than the maximum packet size. It has the characteristic of being. Systematically, the EBRR can be regarded as an extended SRR so that the calculation amount is O (1) even when the quantum value is less than the maximum packet size. The operation of EBRR will be described below.

  EBRR packet scheduling is performed in units of packet flows. That is, in the packet scheduling, a process for determining which flow's leading packet is to be the next packet to be transmitted is performed. The packet flow is specified by, for example, the value of the packet source and destination addresses.

  In the EBRR, for each flow i (i is an identification number for identifying the flow), Queue [i] for storing packets belonging to the flow i as one queue is used for determining whether packets can be transmitted from the flow i. The credit counter Ci, and the credit counter increment value quantum Qi in one scheduling cycle are managed.

  In EBRR, the array ActiveList [] is managed. The index of ActiveList [] corresponds to the scheduling round, and the element of ActiveList [] is a list of Queue [i] of the flow to be scheduled in the scheduling round represented by the index. In the following, the maximum packet size is represented by L, the size of the array ActiveList [] is represented by q, and the current scheduling round is represented by RC (RC = 0, 1, 2,...). At this time, q = ceil (L / (minimum value of Qi)). Note that ceil (x) is a function that returns the smallest integer greater than or equal to x.

The processing in one scheduling round is as follows.
i) The first element (Queue [i]) is extracted from ActiveList [RC mod q] (x mod y indicates a remainder obtained by dividing x by y). One packet is dequeued from the head of Queue [i] and transmitted. Ci is subtracted by the packet length.
ii) If Ci> 0 and Queue [i] is not empty, add Queue [i] to the end of ActiveList [RC mod q]. If Ci ≦ 0, Qi is added to Ci for each scheduling round to obtain a scheduling round rc in which Ci is greater than 0. That is, rc is obtained by setting rc = RC + floor (−Ci / Qi) +1. Note that floor (x) is a function that returns the largest integer less than or equal to x. If Queue [i] is not empty, Queue [i] is added to the end of ActiveList [rc mod q], and Ci is set to Ci (Ci (before update) + (floor (−Ci / Update to the value of Qi) +1) × Qi). If Queue [i] is empty and the packet of Queue [i] is after the scheduling round rc, it is recorded that transmission is possible, and the value of Ci is updated to Qi.
iii) When a packet is received in the scheduling round RC for the flow i, the packet is enqueued in Queue [i]. If Queue [i] is empty before enqueue, and rc is a scheduling round capable of transmitting packets in Queue [i], Queue [i] is ActiveList [max (rc, RC) mod q ] At the end of Note that max (x, y) is a function that outputs the larger of x and y.

L. Lenzini, E.M. Mingozzi and G. Stea, “Eligibility-Based Round Robin for Fair and Efficient Packet Scheduling in Wormhole Switching Networks”, Trans. On Parallel and Distributed Systems, Vol. 15, No. Three

  According to the above-described conventional EBRR technique, it is possible to improve the fairness between flows and reduce the burstiness of flows by reducing the quantum value without changing the ratio between the quantum values of the flows. On the other hand, when the quantum value is decreased, the size (= q) of ActiveList [] increases, and the amount of calculation for checking whether each element of ActiveList [] is empty is increased. Considering that the quantum value represents the amount of data transmitted in one scheduling round, the number of scheduling rounds executed per unit time increases when a smaller data value is used and a constant data transmission rate is maintained. For these reasons, the calculation amount increases when the quantum value is reduced.

  When the scheduling process by EBRR is executed by software using a CPU (Central Processing Unit), the fairness and burstiness are improved according to the load of other processes executed by the CPU, and the calculation amount is not increased. Which one has priority is changed. Therefore, it is desirable to change the quantum value without changing the ratio between the flows in accordance with the load of other processing executed by the CPU. However, in the conventional EBRR, the handling of the transmission waiting packet and data structure when the quantum value is changed during the execution of the scheduling process is not stipulated, and the quantum value is changed according to the load of other processes executed by the CPU. There was a problem that the feasibility of doing was not shown.

  The present invention has been made in view of the above, and when the scheduling process is realized by software by a CPU or the like, the ratio between the flows is set in accordance with the load status of other processes executed by the CPU or the like. It is an object of the present invention to obtain a packet scheduling method and a communication apparatus that can change a quantum value without changing it.

In order to solve the above-described problem and achieve the object, the present invention provides a packet scheduling method for determining a packet to be transmitted for each schedule round using an EBRR scheme in a communication apparatus that performs packet transmission, and , An identification number for identifying a flow, a quantum value that is an increment value per scheduling land of a credit counter in the EBRR system is Qi, j is an integer of 0 or more, and the value of j is the same between flows at the same time. , Qi is a quantum value setting step for setting the minimum quantum value of flow i by 2 ^j , and the maximum value of j is N, and an element of ActiveList [] which is a list of Queue [i] in the EBRR method the stride is the increment per scheduling lands index value to be used when checking the 2 ^j, most of the flow When the minimum value of the quantum value is MinQsys, array size to the size q of the array ActiveList [], q = ceil ( (ceil (L / minQsys) +1) / (2 N)) * (2 N) According to the setting step and the EBRR method, the packet requested for transmission is stored in Queue [i] which is a packet data storage memory queue corresponding to the packet of flow i, and Queue [i] is added to the array ActiveList [] And when the current scheduling land is RC, if Queue [i] is less than 1 and Queue [i] is not empty after sending Queue [i] in the schedule round, Queue [i] ActiveList [(RC + ceil (subject to the necessary credit value ^{/ Qi) * 2 j) mod} q] and the last list to add to the additional step of, when reducing the quantum value, the value of j for a value j far Ri and smaller new value, using the new value of j, include the updating step of updating the stride value to 2 ^j updates the Qi to a value obtained by multiplying a 2 ^j to the minimum quantum value of the flow i It is characterized by.

  According to the present invention, when scheduling processing is realized by software by a CPU or the like, the quantum value is changed without changing the ratio between the flows in accordance with the load status of other processing executed by the CPU or the like. There is an effect that becomes possible.

FIG. 1 is a diagram illustrating a configuration example of a communication apparatus according to the present invention. FIG. 2 is a diagram illustrating an example of a packet transmission processing procedure. FIG. 3A is a diagram illustrating an example of an algorithm for executing the scheduling process. FIG. 3-2 is a diagram illustrating an example of an algorithm for executing the scheduling process. FIG. 3-3 is a diagram illustrating an example of an algorithm for executing the scheduling process. FIG. 3-4 is a diagram illustrating an example of an algorithm for executing the scheduling process. FIG. 3-5 is a diagram illustrating an example of an algorithm for executing the scheduling process.

  Embodiments of a packet scheduling method and a communication apparatus according to the present invention will be described below in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

Embodiment.
FIG. 1 is a diagram illustrating a configuration example of a communication apparatus according to the present invention. The communication device 1 according to the present embodiment includes a control unit 2 such as a CPU or MPU (Micro Processing Unit), a timer unit 3, a credit counter storage memory 4, a quantum (quantum) value storage memory 5, The memory includes an ActiveList storage memory 6, a packet data storage memory 7, a process waiting packet information storage memory 8, a queue information storage memory 9, and a packet input unit 10.

  The communication apparatus according to the present embodiment performs packet scheduling processing by the packet scheduling method according to the present invention (scheduling method that extends EBRR) and transmits the packet. Hereinafter, in the present embodiment, as in normal EBRR, for each flow i (i is an identification number for identifying the flow), the queue for storing packets belonging to the flow i is Queue [i], and the flow The Credit Counter used for determining whether or not to transmit a packet from i is Ci, and the increment value quantum value of the Credit Counter in one scheduling cycle is Qi, and the array ActiveList [] is managed. The index of ActiveList [] corresponds to the scheduling round, and the element of ActiveList [] is a list of Queue [i] of the flow to be scheduled in the scheduling round represented by the index. In the following, the size of the array ActiveList [] is represented by q, and the current scheduling round is represented by RC (RC = 0, 1, 2,...).

  The control unit 2 includes a packet transmission request reception processing unit 21, a scheduling processing unit 22, and a packet transmission processing unit 23. The packet transmission request reception processing unit 21, the scheduling processing unit 22, and the packet transmission processing unit 23 are each stored in a storage unit (not shown) as a software program, and are executed by the control unit 2 according to the program stored in the storage unit when executing processing. Execute each process.

  The packet scheduler of the present embodiment includes a packet transmission request reception processing unit 21, a scheduling processing unit 22, a packet transmission processing unit 23, a timer unit 3, a credit counter storage memory 4, a quantum value storage memory 5, and an ActiveList storage. A memory 6 for storing packet data, a memory 8 for storing waiting packet information, a memory 9 for storing queue information, and a packet input unit 10.

  The packet transmission request reception processing unit 21 performs processing for receiving a packet transmission request from the packet input unit 10 in the communication device 1. The packet that has received the transmission request is stored in the packet data storage memory 7, and information about the packet is stored in the processing-waiting packet information storage memory 8.

  The packet input unit 10 inputs a packet generated by itself (or a packet received from another device or another functional unit) and a transmission request for requesting transmission start of the packet to the packet transmission request reception processing unit 21 of the packet scheduler. To do.

  The scheduling processing unit 22 is activated from the timer unit 3 at a fixed period, performs packet scheduling processing at the activated timing, determines a packet to be transmitted, and outputs a transmission request for the packet to the packet transmission processing unit 23.

  Based on the transmission request received from the scheduling processing unit 22, the packet transmission processing unit 23 performs a process of reading the transmission target packet from the packet data storage memory 7 and transmitting it to the network interface or the like.

  The timer unit 3 manages the cycle for starting the packet scheduling process, and generates an interrupt to the CPU for starting the scheduling processing unit 22 every specified time (interrupt generation cycle). The interrupt generation cycle is designated by the scheduling processing unit 22.

  The packet data storage memory 7 is a memory for storing the packet data body. The queue information storage memory 9 is a memory for storing information of Queue [i] used in the packet scheduling process of the present embodiment. The process waiting packet information storage memory 8 is a memory for storing information on packets waiting to be processed by the scheduling processing unit 22 stored in the packet data storage memory 7 via the packet transmission request reception processing unit 21.

  The credit counter storage memory 4 is a memory for storing information on Ci that is a credit counter used in the packet scheduling processing of the present embodiment. The Quantum value storage memory 5 is a memory that stores information on Qi that is a Quantum used in the packet scheduling process of the present embodiment. The ActiveList storage memory 6 is a memory for storing information of ActiveList [] used in the packet scheduling process of the present embodiment.

  FIG. 2 is a diagram illustrating an example of a packet transmission processing procedure in the communication apparatus according to the present embodiment. A packet transmission process in the communication apparatus according to the present embodiment will be described with reference to FIG. FIG. 2 shows a normal scheduling process without changing the quantum value.

In the present embodiment, when the minimum quantum value of flow i is MinQi, Qi that is the quantum value of flow i is MinQi * 2 ^j (j is an integer of 0 or more). At the same time, the j value of each flow is the same. In this embodiment, an index increment value (RC increment value for each schedule round) used when checking an element of ActiveList [] is called stride, and 2 ^j is used as stride in a normal scheduling process. . Further, when the maximum value that j can take is N, the minimum value among the minimum quantum values for each flow is expressed as minQsys, and the maximum packet length is L, the size q of the array ActiveList [] is
q = ceil ((ceil (L / minQsys) +1) / (2 ^N )) * (2 ^N )
And Further, when Queue [i] is not empty after transmission of a packet in the current scheduling round and Ci is smaller than 0, Queue [i] is ActiveList [(RC + ceil (credit value / Qi that needs addition) * 2 ^j ) Added to the end of mod q]. The credit value that needs to be added is -Ci when Ci is smaller than 0. The process in one scheduling round in the normal scheduling process is the same as the conventional EBRR except for the points described above and the process related to the change of the quantum value described later.

  First, the packet input unit 10 in the communication device 1 inputs a packet transmission request to the packet scheduler (step S1). The packet transmission request acceptance processing unit 21 accepts a packet transmission request, stores the packet data body in the packet data storage memory, and information on the packet (for example, packet length, packet belongs to the processing-waiting packet information storage memory 8) Header information necessary for flow determination) is added to the already stored information and stored (step S2).

  Next, the timer unit 3 manages whether or not a certain time has elapsed (step S3), and when the certain time has elapsed (step S3 Yes), the control unit 2 generates an interrupt to generate a scheduling processing unit. The process 22 is started and the scheduling process is started (step S4). The scheduling processing unit 22 accesses the processing-waiting packet information storage memory 8 and enqueues the packet for each flow by associating the packet data with the flow by the packet arrival processing described later based on the processing-waiting packet information (step S5). . The queue (Queue [i]) corresponding to each flow may be provided in the packet data storage memory 7 or may be provided as a separate queue unit. Thereafter, the scheduling processing unit 22 performs EBRR scheduling processing according to the present embodiment, which will be described later (step S6). The transmission target packet is determined based on the result of the EBRR scheduling processing, and the value of stride at that time. CurStride is determined (step S7).

  The scheduling processing unit 22 notifies the transmission target packet to the packet transmission processing unit 23, and the packet transmission processing unit 23 reads the notified transmission target packet from the packet data storage memory 7 and transmits it (step S8). Next, the scheduling processing unit 22 adds CurStride to RC to update the value of RC, ends the scheduling round processing, and returns to step S3. As described above, the packet scheduler repeats the packet scheduling process at regular intervals.

  In FIG. 1, the configuration unit of the packet scheduler according to the present embodiment is shown as the control unit 2, but actually, the control unit 2 often performs other processes besides the packet scheduling. In EBRR, it is possible to improve the fairness between flows and reduce the burstiness of flows by reducing the quantum value without changing the ratio between the quantum values of the flows. On the other hand, reducing the quantum value increases the amount of calculation. For example, when the load due to other processing is large in the control unit 2, an increase in calculation amount is not desirable, but when the load due to other processing is small in the control unit 2, even if the calculation amount increases, It may be better to prioritize improving fairness and reducing flow burstiness. Thus, the optimum value of the quantum value changes depending on the load caused by other processing in the control unit 2.

In the present embodiment, according to the following procedure, the quantum value can be changed without changing the ratio between the flow quantum values during execution of the scheduling process. As described above, in this embodiment, it has a MinQi * 2 ^j and Qi, when changing the quantum value changes the value of j.

When the quantum value is reduced, j is determined as a value smaller than the previous j, and a new j (or new stride 2 ^j ) is designated to the scheduling processing unit 22 to update the quantum value. Instruct. An instruction to update the quantum value may be automatically generated by a processing unit in the communication device (not shown) according to the load of the control unit 2 or may be input to the scheduling processing unit 22 in response to an input from an operator or the like. It may be input.

When there is an instruction to reduce the quantum value, the scheduling processing unit 22 updates Qi to a value (MinQi * 2 ^j ) corresponding to the new j. Also, the stride value is updated to a value of 2 ^j corresponding to the new j. If the quantum value is decreased and the new 2 ^j value is smaller than the old (previous) 2 ^j value, the old stride value (old 2 ^j value) is a multiple of the new 2 ^j . For this reason, the elements of ActiveList [] that are checked using the new 2 ^j as the stride value include the elements of ActiveList [] that are accessed with the old stride value. Therefore, the stride value can be immediately updated to a new value of 2 ^j (in a scheduling round in which there has been an instruction to reduce the quantum value).

When the quantum value is increased, j is determined as a value larger than the previous j, and a new j (or new stride 2 ^j ) is specified to the scheduling processing unit 22 to update the quantum value. Instruct.

When there is an instruction to increase the quantum value, the scheduling processing unit 22 immediately updates Qi to a value corresponding to the new j (MinQi * 2 ^j ) (in the scheduling round in which there is an instruction to decrease the quantum value). To do. Then, the scheduling processing unit 22 records the index value of the ActiveList [] when the quantum value is changed, and until the index value of the ActiveList [] accessed by the scheduling process becomes equal to the recorded index value, Continue to use the old stride value. The following formula (1) is used when calculating a scheduling round (index value of ActiveList []) that enables packet transmission from Queue [i] after transmitting the packet from Queue [i].
index = ((floor ((RC-1) / new 2 ^j ) +1) * (new 2 ^j ) +
ceil (Credit value that needs addition / Qi) * (new 2 ^j )) mod q (1)

quantum value is incremented, if the value of the new 2 ^j is greater than the value of the old 2 ^j, the elements of the old stride values for all has been accessed (the old value of 2 ^j) ActiveList [] is access once Until then, the old stride value remains as it is and special processing is required as described above. The first part of the above equation (1) obtains the current index value when rounded by the new 2 ^j value so that the index value is a multiple of the new 2 ^j . In this way, the amount of increase in Credit Counter can be made substantially the same for each flow at different scheduling intervals of the same length.

  FIGS. 3-1 to 3-5 are diagrams illustrating an example of an algorithm for executing the scheduling process of the present embodiment. FIGS. 3-1 to 3-5 are examples, and the algorithm for executing the scheduling process according to the present embodiment is an example of FIGS. 3-1 to 3-5 as long as the algorithm can execute the above-described scheduling process. It is not limited.

  This is SystemInit () initialization processing in FIGS. 3-1 to 3-5, and is executed at the start of packet transmission processing. maxExponent corresponds to the above-described N (the maximum value of j), curExponent corresponds to the above-described j, minQuantum corresponds to the above-described minQsys, and macPacketSize corresponds to the above-described L. FlowInit () is executed when a new flow occurs.

ChangeQuantum () is executed when there is an instruction to change the quantum value. newStride corresponds to the new 2 ^j . NeedToStartAdjust is a variable that indicates whether or not to execute the above-mentioned special processing when there is a change that increases the quantum value. NeedToStartAdjust is an initial value of 0 and NeedToStartAdjust is 1 (newStride is from CurStride) Set to 1 if large) to indicate that special processing is to be performed. AdjustInProgress is a variable that indicates whether or not the above-mentioned special processing is being performed when there is a change that increases the quantum value. The initial value is 0, and when AdjustInProgress is 1, special processing is performed. Indicates that it is in progress.

  PacketArrival () is a process corresponding to the packet arrival process described with reference to FIG. 2, and is the same as the conventional ECRR.

  EBRRScheduler2 () is a process corresponding to the EBRR scheduling process described with reference to FIG. 2, and corresponds to a process extended from the conventional EBRR. The special processing described above is performed in EBRRScheduler2 (). NewRound () is a process for updating RC.

As described above, in this embodiment, Qi is set to MinQi * 2 ^j, and stride that is an increment value of an index used when checking an element of ActiveList [] is set to 2 ^j . When the quantum value is decreased, the value of j is decreased, and Qi and stride are immediately updated to values corresponding to the new j. When the quantum value is increased, the value of j is increased, and Qi is changed to the new j. Updates to the corresponding value immediately, does not update stride immediately, records the ActiveList [] index value when the quantum value is changed, and records the ActiveList [] index value accessed in the scheduling process The stride is updated after the index value is reached. For this reason, when the scheduling process is realized by software by a control unit such as a CPU, the quantum value can be changed without changing the ratio between the flows in accordance with the load status of other processes executed by the control unit. It becomes possible.

  Further, among packets received as a plurality of flows in the same scheduling round where the queue is empty and transmission is possible (Credit Counter is a positive value), packets having a small packet size can be transmitted first.

  As described above, the packet scheduling method and the communication apparatus according to the present invention are useful for a communication apparatus that performs packet scheduling for determining a packet to be transmitted at a certain point in time, and particularly suitable for a communication apparatus that executes packet scheduling by software. ing.

DESCRIPTION OF SYMBOLS 1 Communication apparatus 2 Control part 3 Timer part 4 Credit Counter storage memory 5 Quantum (quantum) value storage memory 6 ActiveList storage memory 7 Packet data storage memory 8 Processing waiting packet information storage memory 9 Queue information storage memory 10 Packet input unit 21 Packet transmission request reception processing unit 22 Scheduling processing unit 23 Packet transmission processing unit

Claims (4)

A packet scheduling method for determining a packet to be transmitted for each schedule round using an EBRR method in a communication apparatus that performs packet transmission,
i is an identification number for identifying a flow, a quantum value that is an increment value per scheduling land of a Credit Counter in the EBRR method is Qi, j is an integer of 0 or more, and the value of j is the same between flows at the same time. A quantum value setting step for setting Qi as a value obtained by multiplying the minimum quantum value of flow i by 2 ^j ,
The maximum value of j is N, and the stride that is the increment value per scheduling land of the index value used when checking the elements of ActiveList [] that is the list of Queue [i] in the EBRR method is 2 ^j , When the minimum value of the minimum quantum values of the flow is minQsys, the size q of the array ActiveList [] is q = ceil ((ceil (L / minQsys) +1) / (2 ^N )) * (2 ^N ) An array size setting step,
A data storage step of storing a packet requested for transmission in Queue [i], which is a queue of a packet data storage memory corresponding to the packet of flow i, and adding Queue [i] to the array ActiveList [] according to the EBRR method When,
When RC is the current scheduling land, if Credit Counter is less than 1 and Queue [i] is not empty after transmission of Queue [i] in the schedule round, Queue [i] is set to ActiveList [(RC + ceil (addition A credit value / Qi) * 2 ^j ) mod q] to add to the end of the list addition step;
When the quantum value is reduced, the value of j is changed to a new value smaller than the previous value of j, and Qi is updated to the value obtained by multiplying the minimum quantum value of flow i by 2 ^j using the new value of ^j. And updating the stride value to 2 ^j ,
A packet scheduling method comprising: When the quantum value is increased, the value of j is set to a new value larger than the value of j so far, and the new value of j is used to update Qi to the value obtained by multiplying the minimum quantum value of flow i by 2 ^j. And a quantum update step for storing an index value of ActiveList [] based on a new value of j;
When increasing the quantum value, if the index value of ActiveList [] accessed in the current scheduling round becomes equal to the stored index value, the stride value is updated to 2 ^j based on the new j. A stride update step for updating the index value of ActiveList [] using the stride before the update every schedule round;
The packet scheduling method according to claim 1, further comprising: A communication device that determines a packet to be transmitted for each schedule round using an EBRR method,
i is an identification number for identifying a flow, and a queue unit that includes a Queue [i] for storing a packet for which transmission is requested in a packet of the flow i for each flow;
In the EBRR system, the credit counter increment value per scheduling land is Qi, j is an integer greater than or equal to 0, j is the same between flows at the same time, and Qi is the minimum quantum value of flow i. to set as a value obtained by multiplying the 2 ^j, the maximum value of j is N, a list of Queue [i] in EBRR method ActiveList [] elements per scheduling lands index value to be used when checking the When the stride that is the increment value is 2 ^j and the minimum value of the minimum quantum values of each flow is minQsys, the size q of the array ActiveList [] is set to q = ceil ((ceil (L / minQsys) +1) / (2 ^N )) * (2 ^N ), and according to the EBRR method, a packet for which a transmission request has been made is queued in a queue for storing packet data corresponding to the packet of the flow i. i] and adding Queue [i] to the array ActiveList [] and setting the current scheduling land as RC, the Credit Counter is less than 1 after transmission of Queue [i] in the schedule round, and the Queue If [i] is not empty, the scheduling processing unit adds Queue [i] to the end of ActiveList [(RC + ceil (credit value that needs to be added / Qi) * 2 ^j ) mod q];
With
Wherein the scheduling processing unit, to reduce the quantum value, the value of j to a value smaller than the new value of j so far, by using the new value of j, 2 a Qi to the minimum quantum value of the flow i ^j communication device and updates the stride value to 2 ^j and updates the value obtained by multiplying. Wherein the scheduling processing unit, when increasing the quantum value, the value of j to a value greater than the new value of j so far, by using the new value of j, 2 a Qi to the minimum quantum value of the flow i ^j And the index value of ActiveList [] based on the new j value is stored, and the index value of ActiveList [] accessed in the current scheduling round becomes equal to the stored index value. The stride value is updated to 2 ^j based on a new j, and the index value of ActiveList [] is updated for each schedule round using the stride before the update until stride is updated. Communication device.

Images