WO2003009634A1 - Asymmetric void filling scheduler with bandwidth grabbing - Google Patents

Asymmetric void filling scheduler with bandwidth grabbing Download PDF

Info

Publication number
WO2003009634A1
WO2003009634A1 PCT/US2001/016529 US0116529W WO03009634A1 WO 2003009634 A1 WO2003009634 A1 WO 2003009634A1 US 0116529 W US0116529 W US 0116529W WO 03009634 A1 WO03009634 A1 WO 03009634A1
Authority
WO
WIPO (PCT)
Prior art keywords
schedulers
buffer
scheduler
packets
wavelengths
Prior art date
Application number
PCT/US2001/016529
Other languages
French (fr)
Inventor
Ljubisa Tancevski
Original Assignee
Alcatel Usa Sourcing, L.P.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Alcatel Usa Sourcing, L.P. filed Critical Alcatel Usa Sourcing, L.P.
Priority to PCT/US2001/016529 priority Critical patent/WO2003009634A1/en
Publication of WO2003009634A1 publication Critical patent/WO2003009634A1/en

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04Q—SELECTING
    • H04Q11/00—Selecting arrangements for multiplex systems
    • H04Q11/0001—Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0062—Network aspects
    • H04Q11/0066—Provisions for optical burst or packet networks
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04Q—SELECTING
    • H04Q11/00—Selecting arrangements for multiplex systems
    • H04Q11/0001—Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005—Switch and router aspects
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04Q—SELECTING
    • H04Q11/00—Selecting arrangements for multiplex systems
    • H04Q11/0001—Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005—Switch and router aspects
    • H04Q2011/0007—Construction
    • H04Q2011/002—Construction using optical delay lines or optical buffers or optical recirculation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04Q—SELECTING
    • H04Q11/00—Selecting arrangements for multiplex systems
    • H04Q11/0001—Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005—Switch and router aspects
    • H04Q2011/0037—Operation
    • H04Q2011/005—Arbitration and scheduling
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04Q—SELECTING
    • H04Q11/00—Selecting arrangements for multiplex systems
    • H04Q11/0001—Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0062—Network aspects
    • H04Q2011/0064—Arbitration, scheduling or medium access control aspects

Definitions

  • Figure 2 is a flow diagram, which illustrates the use of virtual wavelength fields and a buffer threshold mechanism.
  • the present invention is an apparatus and method for scheduling packet flow. Its purpose is to reduce excess loads, while at the same time improving efficiency of the system.
  • it uses fiber delay line optical buffers.
  • it can be used to schedule asynchronous, variable length traffic.
  • it can be used under any traffic condition in connection with non-degenerate buffers.
  • the invention is divided into two or more schedulers operating in parallel. Furthermore, each scheduler is assigned a subset of the available wavelengths. The assignment of wavelengths is done in an asymmetric manner such that one scheduler is assigned many more wavelengths. As a result, it experiences a much smaller load (and associated excess load) than the other schedulers. In another preferred embodiment, asymmetric assignment of wavelengths is used in conjunction with bandwidth grabbing to improve scheduling performance.
  • a packet divider/classifier circuit 106 is used to route incoming packets to the appropriate scheduler.
  • a plurality of input fibers 103 are operably connected to the input 105 of the packet divider/classifier 106 and a plurality of schedulers operably connected to the output 107 of the packet divider/classifier circuit 106.
  • One routing approach comprises assigning packets between the two schedulers in the same percentage as the wavelengths that were assigned.
  • buffer occupancy is monitored by buffer monitoring circuitry 108 (or a buffer threshold mechanism) which monitors the occupancy of every virtual wavelength field (see Fig. lb).
  • the maximum occupancy is given by the maximum holding time of the fiber delay line buffer, i.e., the maximum length of the fiber delay line in the buffer.
  • Setting a bandwidth threshold Tl of 0.5 means that there is at least one virtual wavelength field whose occupancy is less than 0.5 of the maximum holding time.
  • bandwidth grabbing or dynamic bandwidth allocation
  • the packet is directed to the wavelength field with minimum occupancy.
  • buffer occupancy monitoring 210 (or buffer monitoring circuitry 108) first determines whether the buffer in scheduler II is fully occupied 212. If scheduler II is not fully occupied 214, the packet is scheduled in a conventional way using the void filling scheduler 216. If scheduler II is fully occupied 218, then buffer monitoring circuitry determines whether the scheduler I is below a threshold 220. If scheduler I is not below a threshold 222, i.e., it is fully occupied, the packet stays in scheduler II. If scheduler I is below a threshold 224, bandwidth grabbing (or dynamic bandwidth allocation) is initiated. The packet is being re-directed to scheduler I and the void filling algorithm of scheduler 1 226 ' controls routing of the packet.

Abstract

The invention is an apparatus and method for scheduling the flow of packets. Its purpose is to reduce excess loads, while at the same time, improving efficiency of the system. The scheduler is divided into two or more schedulers operating in parallel. Furthermore, each scheduler is assigned a subset of the available wavelengths. The assignment of wavelengths is done in an asymmetric manner such that one scheduler is assigned many more wavelengths. In addition, the void filling scheduler is used in conjunction with bandwidth grabbing to improve scheduling performance. In a preferred embodiment, the buffer occupancy of a first scheduler is monitored. If it is below a threshold (and a second scheduler's buffer is full), then packets intended for the second scheduler are transferred to the first scheduler.

Description

ASYMMETRIC VOID FILLING SCHEDULER WITH BANDWIDTH
GRABBING
FIELD OF INVENTION
This invention is related to the field of apparatuses and methods for scheduling the flow of packets. More particularly, it relates to the use of a scheduler for filling voids.
BACKGROUND OF INVENTION
Presently, void filling schedulers utilize only one scheduler operating over the entire set of available wavelengths. (A void is a gap or idle period/pattern of determined time length, appearing at the distribution of packets at the output fields of a switch/router. It essentially represents an absence of packets at the output). Assignment of wavelengths is done in the following manner. Assume that there are N wavelengths available on a certain link. Next, assume that the traffic load per wavelength and traffic overall is p. In addition, assume that there is one scheduler. The scheduler will schedule use of the packets over all N wavelengths, in effect performing interleaving over all the possible wavelengths. This is depicted in Fig. la where it is visible that the scheduler operates over all the wavelengths.
Another example of a void filling scheduler is disclosed in U.S. patent application 09/253,309, Optical Fiber-Delay Line Buffers With Void Filling, filed February 19, 1999 and hereby incorporated by reference. It discloses the use of fiber-delay line buffers that incorporate a void filling procedure.
However, to substantially reduce excess load and increase efficiency using present void filling schedulers, large electronic memories are needed to store large numbers of existing voids ready to be filled. Hence, increased efficiency can only be obtained at the expense of a large increase in the complexity of the void filling scheduling method. In addition, inspecting large number of voids to find out whether a packet can be scheduled in an existing void is very time consuming. Hence, increasing the efficiency entails slowing the speed of the scheduler. Therefore, there is a trade off between complexity and efficiency when using only one scheduler.
SUMMARY OF THE INVENTION
The present invention is a method and apparatus for scheduling packets, comprising a plurality of schedulers connected in parallel and a packet divider/classifier circuit operably connected to the plurality of schedulers.
In another embodiment, the invention is an apparatus for scheduling packets, comprising a packet divider/classifier circuit, a plurality of input fibers operably connected to an input of the packet divider/classifier; and a plurality of schedulers connected in parallel. The plurality of schedulers are operably connected to an output of the packet divider/classifier circuit. Furthermore, at least one of the plurality of schedulers comprises at least one buffer. In addition, buffer monitor circuitry is operably connected to the plurality of schedulers. In a preferred embodiment, the buffers are fiber-delay line buffers.
In still another embodiment, the invention comprises a method of asymmetrically allocating bandwidth, comprising the steps of dividing a bandwidth into available wavelengths and assigning the available wavelengths to a plurality of schedulers, wherein at least one of the plurality of schedulers is assigned a different number of wavelengths than another of the plurality of schedulers.
In yet still another embodiment, the invention comprises a method and apparatus of bandwidth grabbing or dynamic bandwidth allocation in which buffer occupancy of at least one of a plurality of schedulers is momtored and at least one packet is redirected to a scheduler having a buffer occupancy below a threshold. BRIEF DESCRIPTION OF THE DRAWINGS
Figure la shows an example of one scheduler operating over N wavelengths.
Figure lb shows an example of two schedulers operating over N wavelengths.
Figure 2 is a flow diagram, which illustrates the use of virtual wavelength fields and a buffer threshold mechanism.
Figure 3 is a flow chart of one embodiment of the present invention.
Figure 4 is a plot of generated excess load as a function of input traffic load.
Figure 5 is a plot of the probability of packet loss as a function of bandwidth threshold for packets going through both schedulers.
DETAILED DESCRIPTION OF THE INVENTION
The present invention is an apparatus and method for scheduling packet flow. Its purpose is to reduce excess loads, while at the same time improving efficiency of the system. In a preferred embodiment it uses fiber delay line optical buffers. In another preferred embodiment, it can be used to schedule asynchronous, variable length traffic. In addition, it can be used under any traffic condition in connection with non-degenerate buffers.
In a preferred embodiment, the invention is divided into two or more schedulers operating in parallel. Furthermore, each scheduler is assigned a subset of the available wavelengths. The assignment of wavelengths is done in an asymmetric manner such that one scheduler is assigned many more wavelengths. As a result, it experiences a much smaller load (and associated excess load) than the other schedulers. In another preferred embodiment, asymmetric assignment of wavelengths is used in conjunction with bandwidth grabbing to improve scheduling performance.
For example, assume that there are two schedulers 102 and 104 operating in parallel. Each scheduler is assigned a subset of the N available wavelengths. This is depicted in Fig. lb. In a preferred embodiment, Nl wavelengths are assigned to the first_scheduler 102 and N2 wavelengths are assigned to the second scheduler 104 such that N1+N2=N and Nl ≠ N2. Both schedulers operate in parallel and independently.
Furthermore, a packet divider/classifier circuit 106 is used to route incoming packets to the appropriate scheduler. A plurality of input fibers 103 are operably connected to the input 105 of the packet divider/classifier 106 and a plurality of schedulers operably connected to the output 107 of the packet divider/classifier circuit 106. One routing approach comprises assigning packets between the two schedulers in the same percentage as the wavelengths that were assigned.
For example, assume that there are 30 available wavelengths. If the first scheduler 102 is assigned or operates over 20 wavelengths and the second scheduler 104 is assigned 10 wavelengths (that is, 1 :2 ratio), then the packet divider 106 directs 66% of the packets to the first scheduler 102 and 33% to the second scheduler 104. Consequently, both schedulers will see same traffic load as seen by the incoming link. However, the overall system performance will be poor because the schedulers will operate over smaller subsets of wavelengths and the statistical multiplexing gain stemming from multiplexing over a number of wavelengths will be reduced.
In a preferred embodiment, the packet divider/classifier 106 divides incoming traffic 50/50 between the two schedulers 102 and 104. Furthermore, it assigns wavelengths asymmetrically by assigning 25 wavelengths to the first scheduler 102 and assigns 5 wavelengths to the second scheduler 104. This creates a huge imbalance in the load experienced by each scheduler because the first scheduler 102 is going to see a light load and the second scheduler 104 a heavy load. This will produce a low probability of packet loss in the first scheduler 102 and a higher probability of packet loss in the second scheduler 104.
One way to reduce the probability of packet loss is to dynamically allocate bandwidth between the schedulers. Normally, packets destined for the first scheduler 102 will go there. Furthermore, packets directed toward the second scheduler 104 will go to the second scheduler 104. However, if the buffer occupancy of the second scheduler 104 is full and there is a threat of buffer overflow, then packets may be redirected from the second scheduler 104 to the first scheduler 102.
This may occur if the occupancy of the first scheduler 102 is below a threshold Tl. If it is below threshold Tl (and the buffer occupancy of the second scheduler- 104 is full), then packets intended for the second scheduler 104 are transferred to the first scheduler 102. This is an example of one way bandwidth grabbing or dynamic bandwidth allocation. (In a preferred embodiment, a threshold of 0.5 of the maximum buffer occupancy is used. However, this threshold can be either higher or lower depending on the particular application. In another preferred embodiment, the threshold can fall within a range of approximately 0.1 to 1).
An example of a buffer monitoring or management scheme using optical fiber-delay line buffers is shown in Fig. 2. In a preferred embodiment, it uses wavelength division multiplexing (WDM). It creates virtual wavelength fields for each wavelength available on the output link. These wavelength fields govern the distribution of the packets at the output wavelengths.
In addition, buffer occupancy is monitored by buffer monitoring circuitry 108 (or a buffer threshold mechanism) which monitors the occupancy of every virtual wavelength field (see Fig. lb). The maximum occupancy is given by the maximum holding time of the fiber delay line buffer, i.e., the maximum length of the fiber delay line in the buffer. Setting a bandwidth threshold Tl of 0.5 means that there is at least one virtual wavelength field whose occupancy is less than 0.5 of the maximum holding time. When using a bandwidth threshold Tl of 0.5, bandwidth grabbing (or dynamic bandwidth allocation) occurs when a packet originally intended for the second scheduler 104 is redirected to a wavelength field whose occupancy is less than 0.5 in the first scheduler 102. In case there is more than one wavelength field whose occupancy is less than the threshold (e.g. 0.5), then the packet is directed to the wavelength field with minimum occupancy.
Fig. 2 depicts an embodiment in which a first scheduler 102 is operated over a plurality of available wavelengths, λl to λ4, each with a buffer occupancy Bl through B4 respectively. (In a preferred embodiment, there are 4 available wavelengths. However, the number of wavelengths can be either higher or lower depending on the particular application). The buffer threshold Tl is depicted as being 0.5 of the maximum buffer occupancy, that is 0.5 of the maximum length of the available delay line. In this example, only wavelength field λ4 has a buffer occupancy B4 below the threshold Tl . Consequently, if the second scheduler 104 is 100% occupied, the packet will be redirected to the first scheduler 102 as depicted in Fig. 2 and will be scheduled at λ4. Hence the bandwidth grabbing mechanism has redirected the packet. The situation in Fig. 2 corresponds directly to the situation in the void filling scheduler and, for that reason, existing voids, VI, N2, N3 and N4, are also indicated.
A flow chart of the present invention is depicted in Fig. 3. Upon arrival, the packet divider 106 senses a new packet arrival 200 and determines to which scheduler the new packet will go 202, i.e., assigns the new packet to a scheduler. This can be done either on the basis of some statistics (such that 50% of the traffic goes to each scheduler) or on basis of class differentiation. (This is done in differentiated service environments). After making the decision, the packet is directed toward the appropriate scheduler.
If the packet is assigned or directed to scheduler 1204, everything is done in a conventional way. If the packet is assigned to scheduler II 208, buffer occupancy monitoring 210 (or buffer monitoring circuitry 108) first determines whether the buffer in scheduler II is fully occupied 212. If scheduler II is not fully occupied 214, the packet is scheduled in a conventional way using the void filling scheduler 216. If scheduler II is fully occupied 218, then buffer monitoring circuitry determines whether the scheduler I is below a threshold 220. If scheduler I is not below a threshold 222, i.e., it is fully occupied, the packet stays in scheduler II. If scheduler I is below a threshold 224, bandwidth grabbing (or dynamic bandwidth allocation) is initiated. The packet is being re-directed to scheduler I and the void filling algorithm of scheduler 1 226'controls routing of the packet.
As a result of using asymmetric bandwidth allocation, scheduler I operates over many more wavelengths. Therefore, it will experience very low traffic load. Furthermore, the amount of created voids and excess load will be extremely low.
Bandwidth grabbing redirects some of the packets from scheduler II to scheduler I, in essence increasing the traffic load experienced by scheduler I and also the number of voids. However, because creation of voids and excess load is a strongly non-linear process, the overall excess load, as seen by both schedulers, will be reduced. Hence, the probability of packet loss using two schedulers (or a plurality of schedulers) will be lower than the probability of packet loss when using only one scheduler. Hence, by using asymmetric bandwidth allocation in connection with bandwidth grabbing (or dynamic bandwidth allocation) it is possible to increase the efficiency of the void filling method to fight the occurrence of the excess load. Therefore, the proposed apparatus and method reduces the overall excess load and increases efficiency without escalating complexity.
To evaluate the dependence of the excess load on the input traffic load, a test was run for the case of asynchronous, fixed length bursty traffic with a burstiness of 4. The packet size was set at 400 bytes, the selected basic delay line in the buffer was set at 2000 bytes and buffer depth was set at 6. It was assumed that the system contained a 4X4 non- blocking switch and 16 wavelengths per fiber. The generated excess load as a function of the input traffic load is plotted in Fig. 4. Note that for high input loads the excess load measures as high as 0.2 but for traffic loads below 0.75 it drops to increasingly small values. This is an indication that reducing the traffic load artificially will improve the efficiency of the void filling scheduler. Note the non-linear dependence.
To test the probability of packet loss for a system with more than one scheduler, the following simulation was run. The number of wavelengths per input fiber was fixed at 32 and traffic was divided 50/50 between two schedulers. The input traffic load was fixed at 0.8. The occupancy of the buffers of both schedulers was then monitored; When scheduler II was fully occupied (in a preferred embodiment, it was 100% occupied) and scheduler I's occupancy was below a threshold, bandwidth grabbing was initiated.
The probability of packet loss as a function of the bandwidth threshold for packets going through both schedulers is plotted in Fig. 5. Numbers in circles are used- to indicate bandwidth or wavelength allocation. Dashed lines represent the probability of packet loss for scheduler II and full lines represent the probability of packet loss for scheduler I.
As shown in Figure 5, a threshold Tl of 0.1 means that bandwidth grabbing almost never occurs and a buffer threshold Tl of 1 means that bandwidth grabbing almost always occurs. For thresholds very close to 0.9, both schedulers experience almost the same performance. The stronger the asymmetry of wavelength allocation, the better the overall performance. This is demonstrated by the fact that the 13/19 partition features better performance than the 14/18 partition.
However, both partitions feature much better performance than the case when there is only one scheduler operating over all 32 wavelengths as indicated by straight line 32. Here, in the case of void filling scheduler, the non-linearity of the excess load process assures that the overall performance will be better when dividing the traffic in an asymmetric way between several schedulers operating in parallel.
While the invention has been disclosed in this patent application by reference to the details of preferred embodiments of the invention, it is to be understood that the disclosure is intended in an illustrative rather than in a limiting sense, as it is contemplated that modification will readily occur to those skilled in the art, within the spirit of the invention and the scope of the appended claims and their equivalents.

Claims

What is claimed is :
1) An apparatus for scheduling packets, comprising: a plurality of schedulers; and a packet divider/classifier circuit operably connected to said plurality of schedulers.
2) The apparatus for scheduling packets according to claim 1, wherein said plurality of schedulers are operably connected in parallel.
3) The apparatus according to claim 1, further comprising: buffer monitor circuitry operably connected to said plurality of schedulers; and wherein at least one of said plurality of schedulers comprises at least one buffer.
4) The apparatus according to claim 1 , further comprising a plurality of input fibers operably connected to an input of said packet divider/classifier.
5) The apparatus according to claim 3, wherein said buffer is a fiber-delay line buffer.
6) The apparatus according to claim 1, further comprising: a plurality of input fibers operably connected to an input of said packet divider/classifier; wherein said plurality of schedulers are connected in parallel and at least one of said plurality of schedulers comprises at least one buffer; and buffer monitor circuitry operably connected to said plurality of schedulers.
7) The apparatus according to claim 6, wherein said at least one buffer is a fiber-delay line buffer.
8) A method of scheduling packets, comprising: dividing a bandwidth into available wavelengths; and assigning said available wavelengths to a plurality of schedulers.
9) The method according to claim 8, wherein said assignment of wavelengths is done in an asymmetric manner, whereby at least one of said plurality of schedulers is assigned a different number of wavelengths than another of said plurality of schedulers.
10) The method according to claim 8, further comprising the step of assigning packets between said plurality of schedulers in a same percentage as the wavelengths that were assigned.
11) The method according to claim 8, further comprising the step of assigning packets between said plurality of schedulers in a different percentage as the wavelengths that were assigned.
12) The method according to claim 8, further comprising the step of dynamically allocating bandwidth between said plurality of schedulers.
13) The method according to claim 8, further comprising: assigning said available wavelengths to said plurality of schedulers in an asymmetric manner, whereby at least one of said plurality of schedulers is assigned a different number of wavelengths than another of said plurality of schedulers; assigning packets between said plurality of schedulers; and dynamically allocating bandwidth between said plurality of schedulers comprising the steps of : monitoring buffer occupancy of at least one of said plurality of schedulers, determining if a buffer of said at least one of said plurality of schedulers is fully occupied; and redirecting at least one packet to at least one other of said plurality of schedulers having a buffer occupancy below a threshold. 14) The method according to claim 12, wherein said step of dynamically allocating bandwidth comprises: monitoring buffer occupancy of at least one of said plurality of schedulers; and redirecting at least one packet to at least one of said plurality of schedulers having said buffer occupancy below a threshold.
15) The method according to claim 12, said step of dynamically allocating bandwidth comprises: monitoring buffer occupancy of at least one of said plurality of schedulers; determining if a buffer of said at least one of said plurality of schedulers is fully occupied; and redirecting at least one packet to at least one other of said plurality of schedulers having a buffer occupancy below a threshold.
16) The method according to claim 12, wherein said step of dynamically allocating bandwidth comprises: monitoring buffer occupancy of at least one of said plurality of schedulers; determining if a buffer of said at least one of said plurality of schedulers is fully occupied; and redirecting at least one packet to at least one other of said plurality of schedulers having a minimum buffer occupancy.
17) The method according to claim 14, wherein said threshold falls within a range of 0.1 to 1 of said buffer's maximum occupancy.
18) The method according to claim 14, wherein said step of monitoring further comprises monitoring the occupancy of at least one virtual wavelength field.
19) The method according to claim 15, wherein said threshold is equal to or greater than 0.5 of said buffer's maximum holding time. 20) A method of scheduling packets, comprising: receiving packets; dividing said packets into more than one class; and directing said packets toward more than one scheduler.
21) The method according to claim 20, wherein said step of dividing said packets is done statistically.
22) The method according to claim 20, wherein said step of dividing said packets is done on the basis of class differentiation.
23) The method according to claim 20, further comprising the step of dynamically allocating bandwidth between said plurality of schedulers, wherein said step of dynamically allocating bandwidth comprises: monitoring buffer occupancy of at least one of said more than one scheduler; and redirecting at least one packet to a scheduler having at least one buffer with an occupancy below a threshold.
24) The method according to claim 23, wherein said threshold is equal to or greater than 0.5 of said buffer's maximum holding time.
25) A method of bandwidth grabbing between a plurality of schedulers, comprising the steps of :
monitoring buffer occupancy of at least one of said plurality of schedulers; and redirecting at least one packet to a scheduler having a buffer occupancy below a threshold.
26) The method according to claim 25, further comprising the step of determining if a buffer of said at least one of said plurality of schedulers is fully occupied. 27) The method according to claim 25, wherein said step of monitoring further comprises monitoring the occupancy of at least one virtual wavelength field.
28) The method according to claim 25, wherein said threshold is equal to or greater than 0.5 of said buffer's maximum holding time.
PCT/US2001/016529 2001-07-18 2001-07-18 Asymmetric void filling scheduler with bandwidth grabbing WO2003009634A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/US2001/016529 WO2003009634A1 (en) 2001-07-18 2001-07-18 Asymmetric void filling scheduler with bandwidth grabbing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/US2001/016529 WO2003009634A1 (en) 2001-07-18 2001-07-18 Asymmetric void filling scheduler with bandwidth grabbing

Publications (1)

Publication Number Publication Date
WO2003009634A1 true WO2003009634A1 (en) 2003-01-30

Family

ID=21742591

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2001/016529 WO2003009634A1 (en) 2001-07-18 2001-07-18 Asymmetric void filling scheduler with bandwidth grabbing

Country Status (1)

Country Link
WO (1) WO2003009634A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007538374A (en) * 2004-05-17 2007-12-27 トヨタ モーター エンジニアリング アンド マニュファクチャリング ノース アメリカ インコーポレイテッド Batteries with molten salt electrolyte and phosphorus-containing cathode

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2000032008A1 (en) * 1998-11-25 2000-06-02 Italtel S.P.A. Method and device for the wfq statistical multiplexing of atm flows
EP1030534A2 (en) * 1999-02-17 2000-08-23 Alcatel Optical fiber-delay line buffers with void filling
US6169728B1 (en) * 1996-03-29 2001-01-02 Motorola Inc. Apparatus and method for spectrum management in a multipoint communication system
EP1089498A2 (en) * 1999-09-30 2001-04-04 Alcatel Control architecture in optical burst-switched networks

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6169728B1 (en) * 1996-03-29 2001-01-02 Motorola Inc. Apparatus and method for spectrum management in a multipoint communication system
WO2000032008A1 (en) * 1998-11-25 2000-06-02 Italtel S.P.A. Method and device for the wfq statistical multiplexing of atm flows
EP1030534A2 (en) * 1999-02-17 2000-08-23 Alcatel Optical fiber-delay line buffers with void filling
EP1089498A2 (en) * 1999-09-30 2001-04-04 Alcatel Control architecture in optical burst-switched networks

Non-Patent Citations (7)

* Cited by examiner, † Cited by third party
Title
KAZUHIRO OHTSUKI ET AL: "A HIGH-SPEED PACKET SWITCH ARCHITECTURE WITH A MULTICHANNEL BANDWIDTH ALLOCATION", NETWORKING IN THE NINETIES. BAL HARBOUR, APR. 7 - 11, 1991, PROCEEDINGS OF THE ANNUAL JOINT CONFERENCE OF THE COMPUTER AND COMMUNICATIONS SOCIETIES. (INFOCOM), NEW YORK, IEEE, US, vol. 1 CONF. 10, 7 April 1991 (1991-04-07), pages 155 - 162, XP000223501, ISBN: 0-87942-694-2 *
PATTAVINA A: "MULTICHANNEL BANDWIDTH ALLOCATION IN A BROADBAND PACKET SWITCH", IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, IEEE INC. NEW YORK, US, vol. 6, no. 9, 1 December 1988 (1988-12-01), pages 1489 - 1499, XP000001605, ISSN: 0733-8716 *
TANCEVSKI L ET AL: "A NEW SCHEDULING ALGORITHM FOR ASYNCHRONOUS, VARIABLE LENGTH IP TRAFFIC INCORPORATING VOID FILLING", OFC/IOOC '99 OPTICAL FIBER COMMUNICATION CONFERENCE / INTERNATIONAL CONFERENCE ON INTEGRATED OPTICS AND OPTICAL FIBER COMMUNICATION. SAN DIEGO, CA, FEB. 21 - 26, 1999, OPTICAL FIBER COMMUNICATION CONFERENCE / INTERNATIONAL CONFERENCE ON INTEGRATED OP, 21 February 1999 (1999-02-21), pages THM7 - 1-THM7-3, XP000967023, ISBN: 0-7803-5430-3 *
TURNER J S: "WDM BURST SWITCHING FOR PETABIT DATA NETWORKS", OPTICAL FIBER COMMUNICATION CONFERENCE. (OFC). TECHNICAL DIGEST POSTCONFERENCE EDITION. BALTIMORE, MD, MARCH 7 - 10, 2000, NEW YORK, NY: IEEE, US, vol. 2 OF 4, 7 March 2000 (2000-03-07), pages WD2 - 1-WD2-3, XP001017463, ISBN: 0-7803-5952-6 *
TURNER, JONATHAN S., PROCEEDINGS OF INET99, June 1999 (1999-06-01), XP002196966, Retrieved from the Internet <URL:http://www.arl.wustl.edu/~jst/pubs/inet99/inet99.html> [retrieved on 20020418] *
WEN DE ZHONG ET AL: "AN OUTPUT-BUFFERED ATM SWITCH IN THE MULTICHANNEL BANDWIDTH ALLOCATION ENVIRONMENT", ELECTRONICS & COMMUNICATIONS IN JAPAN, PART I - COMMUNICATIONS, SCRIPTA TECHNICA. NEW YORK, US, vol. 74, no. 6, 1 June 1991 (1991-06-01), pages 13 - 23, XP000275277, ISSN: 8756-6621 *
YUHUA CHEN ET AL: "WDM burst switching for petabit capacity routers", MILITARY COMMUNICATIONS CONFERENCE PROCEEDINGS, 1999. MILCOM 1999. IEEE ATLANTIC CITY, NJ, USA 31 OCT.-3 NOV. 1999, PISCATAWAY, NJ, USA,IEEE, US, 31 October 1999 (1999-10-31), pages 968 - 973, XP010369693, ISBN: 0-7803-5538-5 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007538374A (en) * 2004-05-17 2007-12-27 トヨタ モーター エンジニアリング アンド マニュファクチャリング ノース アメリカ インコーポレイテッド Batteries with molten salt electrolyte and phosphorus-containing cathode

Similar Documents

Publication Publication Date Title
US4769811A (en) Packet switching system arranged for congestion control
US5541912A (en) Dynamic queue length thresholds in a shared memory ATM switch
US4769810A (en) Packet switching system arranged for congestion control through bandwidth management
Dittmann et al. Network processor load balancing for high-speed links
US6728212B1 (en) Asymmetric void filling scheduler with bandwidth grabbing
US6014384A (en) Method for controlling data traffic in an ATM network
JP5005048B2 (en) Method for allocating dynamic bandwidth in passive optical networks where different ONUs transmit at different rates
JP3793189B2 (en) Dynamic bandwidth allocation method using tree algorithm and Ethernet (registered trademark) passive optical subscriber network using the same
JPH06209329A (en) Virtual path band assigning system in asynchronous transfer mode
US8000234B2 (en) Dynamic bandwidth allocation based on partial cycles
CN102158420B (en) Service traffic scheduling method based on priority queue and device thereof
Choi et al. Double-phase polling algorithm based on partitioned ONU subgroups for high utilization in EPONs
EP1215931B1 (en) Distributed scheduler for packet switches and passive optical networks
JP3228256B2 (en) Packet communication system, network-side device, and time slot allocation control method
Ma et al. A fairness-aware dynamic spectrum allocation scheme in elastic optical networks
WO2003009634A1 (en) Asymmetric void filling scheduler with bandwidth grabbing
JP3644404B2 (en) Optical subscriber line terminal station apparatus, APON system, and cell delay fluctuation suppressing method
Zang et al. Photonic slot routing in all-optical WDM mesh networks
Ribeiro et al. Improvements on performance of photonic packet switching nodes by priority assignment and buffer sharing
Savi et al. Data-plane architectures for multi-granular OBS network
Veisllari et al. Scheduling techniques in an integrated hybrid node with electronic buffers
JPH0677987A (en) Transfer control system by quality classes for atm exchange
KR100204492B1 (en) Method for ensuring the jitter in hrr queueing service of atm networks
Benzaoui et al. Electronic architectures of optical slot switching nodes
Chan et al. Comparison of static and dynamic bandwidth allocation schemes for multiple QoS classes in ATM networks

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AU CA CN SG

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE TR

121 Ep: the epo has been informed by wipo that ep was designated in this application
DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
122 Ep: pct application non-entry in european phase