WO2005094012A1

WO2005094012A1 - Method and apparatus for overload handling at a processing device

Info

Publication number: WO2005094012A1
Application number: PCT/EP2004/003275
Authority: WO
Inventors: Eckhard Delfs
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2004-03-27
Filing date: 2004-03-27
Publication date: 2005-10-06

Abstract

The invention provides a method and a device for overload handling at a processing device for processing data from two or more channels (C1; C2; C3; C4). The method provided by the invention comprises the steps of: assigning (S 1) to each of the two or more channels first time intervals (TSEI ... TSG1; TSE2 ... TSG2; TSE3 ... TSG3; TSE4 ... TSG4) outside of which processing data of for a respective channel may be selectively denied: selecting (S4) from the two or more channels (C1; C2; C3; C4) a first channel for that processing of data shall be denied; memorizing (S5) the first channel, detecting (S6) expiring of a particular one of first time intervals previously assigned to the first channel, deciding (S7) that the processing of data from the first channel shall be denied, and selecting (S8) a further of the two or more channels (C1; C2; C3; C4) for that processing of data shall be denied.

Description

Method and apparatus for overload handling at a processing device

Field of invention

The invention relates to a method and a device for overload handling at a processing device when applying statistical multiplexing.

Description of prior art

For processing a plurality of channels with a varying demand on a capacity by a resource commonly utilized by the plurality of channels the concept of statistical multiplexing has been introduced. The commonly utilized resource may be for example a common transmission channel on that a plurality of channels is multiplexed. In this context the commonly utilized transmission channel is limited by a maximum bit rate and the varying demand on the resource is the varying bit rate of the different channels that are multiplexed on the common transmission channel.

Another example is a common encoding or decoding device that is utilized by a plurality of speech channels that have a variable demand on the processing capacity of the encoding device respectively the decoding device.

According to the concept of statistical multiplexing the commonly utilized resource is 'overbooked' by the channels utilizing the resource, i.e. the sum of the maximum demands on a capacity of the channels utilizing the resource exceeds the maximum capacity of the commonly utilized resource. By this the fact is utilized that the varying demands on the common resource compensate each other to some extend and that a sum of the capacity demands is less variant than the single capacities.

However even the sum of the capacity demands is time varying and may sometimes exceed the maximum available capacity of the commonly utilized device. This situation is termed an overload situation According to the state of the art the different channels are utilizing a common processing device on a best effort basis, i.e. a channel may utilize the common processing device when it is available and utilization of the processing device may be denied when the maximum capacity is already utilized by other channels.

When scheduling the utilizing of the common processing device by various channels on a best effort basis, the utilizing of the common processing device may be repeatedly denied for a specific channel in an overload situation. By this the transmission quality of said specific channel is severely degraded.

Object of the invention

Therefore it is object of the invention to provide a method and a device for commonly utilizing a common processing device by a plurality of channels, that avoids a severe degrading of the transmission quality.

Summary of the invention

This object is achieved by the method of claim 1 and the scheduling unit of claim 8. Advantageous embodiments are described in the depending claims.

According to the invention a method is provided for overload handling at a processing device for processing data from two or more channels.

The provided method comprises the steps of:

- Assigning to each of the two or more channels first time intervals outside of which processing data of for a respective channel may be selectively denied,

- selecting from the two or more channels a first channel for that processing of data shall be denied,

- memorizing the first channel,

- detecting expiring of a particular one of first time intervals previously assigned to the first channel,

- deciding that the processing of data from the first channel shall be denied, and - selecting a further of the two or more channels for that processing of data shall be denied. By this a denying a processing of data can be controlled and the situation can be avoided that processing of data is repeatedly denied for the same channel.

In an advantageous embodiment of the invented method, it comprises the additional steps of: - assigning to each of the two or more channels second time intervals outside of which processing of data from a respective channel shall be denied independently from the selection of the first or the further channel, said second time intervals enclosing respective first time intervals,

- detecting expiring of a particular one of said second time intervals, and - deciding that processing of data from a channel previously assigned to the particular one of said second time intervals shall be denied.

By this an accumulation of a processing delay can be advantageously avoided.

In a further embodiment of the above methods a bit rate of at least one of the two or more channels is time - variant.

This is an example of a channel having a varying demand on a transmission capacity.

Examples for a kind of channels that are time variant are speech channels.

In another embodiment of the invented method the processing of data comprises an encoding or a decoding of speech data.

Encoding or decoding of speech data are examples for a processing of data that may have a varying demand on a processing capacity or a transmission capacity.

In a further embodiment of the invented method the processing of data from the two or more channels comprises a transmitting of data from the two or more transmission channels via a common transmission channel. In other words the two or more channels are multiplexed on the common transmission channel.

In an embodiment of the invented method the processing of data comprises a processing of data portions and denying processing of data from a determined channel, i.e. the selected or the particular channel, comprises discarding a data portion received on the determined channel.

Examples for data portions are packets for transmitting data or frames for transmitting data, such as speech frames.

Furthermore the invention provides a control unit for overload handling at a processing device for processing data from two or more channels. The control unit may perform any of the above-described methods. The control unit comprises

- a first time table assigning to each of the two or more channels first time intervals outside of which processing of data from a respective channel may be selectively denied, - a selection unit for selecting from the two or more channels a first channel for that processing of data shall be denied,

- a memory unit for memorizing the first channel,

- a timing unit for detecting expiring of a particular one of first time intervals previously assigned to the first channel, and - a decision logic for deciding that the processing of data from the first channel shall be denied, wherein the selection unit is further adapted to select a further of the two or more channels for that the processing of data shall be denied.

The above control unit denies a processing of data in a controlled way and avoids the situation that processing of data is repeatedly denied for the same channel.

In an advantageous embodiment of the control unit it further comprises:

- a second time table defining for each of the two or more channels second time intervals outside of which processing of data shall be denied independently from the selection of the first or the further channel, said second time intervals enclosing respective first time intervals, - a timing unit for detecting expiring of a particular one of said second time intervals, and

- a decision logic for deciding that the utilization of the processing of data shall be denied for a channel previously assigned to the particular one of said second time intervals. With this embodiment of the control unit an accumulation of a processing delay can be advantageously avoided.

Brief description of the drawings

Figure 1 depicts a method for commonly utilizing a processing device by a plurality of channels.

Figure 2 depicts a scheduling table for applying the invention.

Figure 3 depicts a processing device and a control unit receiving data on various channels.

Detailed description of embodiments

Figure 1 depicts steps of a method for overload handling at a processing device according to the invention. In the method of figure 1 a processing device is commonly utilized by a plurality of channels or in other words the processing device is processing data from various channels

The various channels have each a time-variant demand on a capacity of the commonly utilized processing device. The time-variant demand on a capacity can be e.g. a time- variant demand on processing capacity or a transmission capacity. Examples for a kind of channel with a time- variant demand on a processing capacity and/or a transmission capacity are speech channels transmitting encoded speech. In this case the processing device may be an encoder or a decoder commonly utilized by a plurality of speech channels.

The commonly utilized processing device may be a transmission channel on that the various channels are multiplexed.

The various channels may be channels of different kind or channels of the same kind and the processing device may be implemented in hardware or software or in a combination thereof. First time intervals are defined in a first step SI for utilizing the common processing device and assigned to the various channels. One or more of said first time intervals are defined for each of said various channels. On each of said first time intervals processing of data received on a respective channel is scheduled. Outside of the first time intervals utilizing the processing device may be selectively denied for the respective channel. Said first time intervals may e.g. indicate intervals, at which processing of data from a specific channel may start. Processing of data from a specific channel may comprise processing data portions received on the channel. An example for data that is transmitted in data portions is speech data that is transmitted in speech frames.

Said first time intervals may be periodically repeated for each of said channels after a repetition interval that is common to the various channels, e.g. to schedule a processing of succeeding data portions of a respective channel. As will be described in more detail related to figure 2, said first time intervals may partially overlap to provide a statistical multiplexing gain.

The processing of data related to the various channels is scheduled in a third step S3 using the first time intervals defined for the various channels. I.e. the time intervals are used as a scheduling table and starting processing of data received on a specific channel is scheduled within the respective one of said first time intervals.

A particular one of the various channels, termed a first channel, is selected from the multiple channels in a fourth step S4 as the next channel to be skipped i.e. the next channel for that using the commonly utilized processing device shall be denied. The first channel is memorized in a fifth step S5. To that end an indication of the channel may be stored or a counter may be set to a value representing the selected channel.

When scheduling the processing of data from the various channels, an expiring of that one of said first time intervals, on that utilizing the processing device by the selected first channel has been scheduled is detected. This may have occurred when a processing of data received on a previously scheduled channel exceeded the end of the one of the first time intervals assigned to the selected channel within the repetition interval. Detecting expiring of a time interval on that processing of data from a channel has been scheduled is performed in a sixth step S6 and may be termed detecting an overload situation.

Upon detecting expiring of the time interval on that processing of data from the selected channel has been scheduled, the selected channel is skipped in a seventh step S7, i.e. processing of data from the selected channel is denied. This may comprise discarding a data portion received on the selected channel.

Upon skipping the previously selected channel a further channel to be skipped next is selected in an eighth step S8 and advantageously memorized. This can be performed by increasing or decreasing a counter value representing a selected one of the various channels.

Advantageously second time intervals are defined in a second step S2 in addition to the first time intervals, each enclosing a respective one of said first time intervals. Said second time intervals are assigned to each of said various channels. Outside of said second time intervals processing of data from a channel previously assigned to a respective one of second time intervals may be denied independently from a selection of a channel to be skipped.

Like the first time intervals the second time intervals may be periodically repeated after a repetition interval that is common to the various channels. The beginning of each of said second time intervals may be seen as an earliest possible start time to process data from a respective channel within the repetition interval and the end of each of said second time intervals may be seen as a latest possible start time to process data from the respective channel within the repetition interval.

When scheduling the processing of data from the various channels, an expiring of that one of said second time intervals, on that utilizing the processing device by one of said various channels has been scheduled is detected. This may have occurred when a processing of data from previously scheduled channel exceeded the end of that one of second channels. Upon said detecting expiring of the one of said time intervals, the channel previously assigned to the expired time interval is skipped, i.e. processing of data from the previously assigned channel is denied. This may comprise discarding a data portion received on the previously assigned channel.

Figure 2 depicts a scheduling table ST according to the invention for scheduling processing of data by a processing device that is commonly utilized by a plurality of channels. The scheduling table comprises first time intervals outside of which processing of data from a respective channel may be selectively denied as has been described related to figure 1. Further the scheduling table ST comprises second time intervals outside of which processing of data from a respective channel may be denied independently from a previous selection of a channel, i.e. a previous selection of a channel to be skipped.

In the depicted example four channels are commonly utilizing a processing and/or transmission capacity of a processing device. To each of said four channels a respective one of first time intervals is assigned starting at a respective one of a first to fourth earliest starting time TSE 1 ... TSE4 and ending at a respective one of a first to fourth start of guard period TSGl ... TSG4. Said first time intervals may be repeated after a repetition interval RI as is depicted by example of a next earliest start time TSEli for the first channel within the next repetition interval, and a next start of the guard period TSGli for the first channel. The scheduling table ST further comprises first to fourth latest start times TSLl ... TSL4 assigned to a respective one of the four channels, thereby defining second time intervals starting at a respective one of the first to fourth earliest starting time TSEI ... TSE4 and ending at a respective one of the first to fourth latest start time TSLl ... TSL4. As the first time intervals the second time intervals may be repeated after the repetition interval RI.

The sum of the second time intervals may exceed the repetition interval RI, thereby providing to each channel a maximum potential capacity of the processing device that exceeds a nominal share of the processing capacity of the respective device. This is termed a statistical multiplexing gain. Each of the second time intervals encloses a respective one of first time intervals. Intervals within second intervals exceeding respective first intervals are termed guard periods; i.e. a first to fourth guard period is defined starting at a respective one of the first to fourth start of guard period TSGl ... TSG4 and ending at a respective one of the first to fourth latest start time TSLl ... TSL4.

One of first to fourth channels is selected as the next channel to be skipped, i.e. the next channel for that processing of data from it may be denied.

The processing of data received on a specific channel is scheduled to start during a respective one of the second time intervals previously assigned to said specific channel. E.g. the processing of data from the first channel within the repetition interval RI is scheduled between the first earliest start time TSEI and the first latest start time TSLl. If during scheduling processing of data the processing time of a previously scheduled channel exceeded the respective first time interval assigned to a currently scheduled channel, a check is performed, whether the currently scheduled channel is selected as the next channel to be skipped. If the selected channel matches to the currently scheduled channel, the selected channel is skipped, or in other words, processing of data from the selected channel is denied, and data from a next scheduled channel is processed. A different channel than the previously selected channel is selected as the next channel to be skipped. For that a counter value representing a respective selected channel may be increased or decreased.

If during scheduling processing of data the processing time of a previously scheduled channel exceeded the respective second time interval assigned to a currently scheduled channel, the currently scheduled channel is skipped independently from a previous selection of a channel to be skipped. Processing of data from the currently scheduled channel is denied, and data from a next scheduled channel is processed. To distribute a denying a processing of data evenly among the various channels, the currently scheduled channel may be memorized to be selected less often as next channel to be skipped in the future. Figure 3 depicts a control unit CU for overload handling at a processing device PU for processing data from the first to fourth channel CI ... C4. The control unit CU comprises a scheduling table ST, a memory unit MU, a decision logic DL, a timing unit TU, a selection unit SU and a multiplexing unit MD, all of them interconnected by an interface IB providing information exchange between the various units.

The scheduling table ST provides scheduling processing data from the first to fourth channel CI ... C4. The scheduling table ST comprises a first time table assigning to each of the first to fourth channel CI ... C4 first time intervals outside of which processing of data from a respective channel may be selectively denied. The multiplexing unit MD is receiving data on the first to fourth channel C 1 ... C4 and forwards data from a currently scheduled one of said channels to the processing device PU. The selection unit SU provides selecting from the first to fourth channel CI ... C4 a channel for that processing of data shall be denied next. The memory unit MU provides memorizing the selected channel. The memory unit MU can comprise a counter storing a counter value representing a respective one of the first to fourth channel C 1 ... C4.

The timing unit TU provides detecting expiring of a particular one of first time intervals previously assigned to the currently selected channel. To that end the timing unit TU comprises a clocking device and the timing unit TU uses information provided by the scheduling table ST. The decision logic DL provides deciding that the processing of data from the currently selected channel shall be denied. To that end an indication from the timing unit may be used. According to the decision within the decision logic DL, the multiplexing unit MD is indicated that processing data from the currently selected channel shall be denied. Accordingly the multiplexing unit MD switches to the next scheduled channel and forwards data from the next scheduled channel to the processing device PU. To determine the next scheduled channel information from the scheduling table ST can be used. The selection unit SU selects a different channel than the previously selected channel and indicates the newly selected channel towards the memory unit MU that memorizes the newly selected channel.

The scheduling table ST advantageously comprises also a second time table defining for each of the first to fourth channel CI ... C4 second time intervals outside of which processing of data shall be denied independently from a selection of a channel. In that case the timing unit TU additionally provides detecting expiring of a particular one of said second time intervals, and the decision logic DL further provides deciding that the utilization of the processing of data shall be denied for the a channel previously assigned to that particular one of said second time intervals.

According to the decision within the decision logic DL, the multiplexing unit MD is indicated that processing data from the channel previously assigned to said expired time interval shall be denied. Accordingly the multiplexing unit MD switches to the next scheduled channel and forwards data from the next scheduled channel to the processing unit. To determine the next scheduled channel information from the scheduling table ST can be used.

Claims

1. Method for overload handling at a processing device for processing data from two or more channels (CI; C2; C3; C4), the method comprising the steps of: - assigning (SI) to each of the two or more channels first time intervals (TSEI ... TSGl; TSE2 ... TSG2; TSE3 ... TSG3; TSE4 ... TSG4) outside of which processing data of for a respective channel may be selectively denied, - selecting (S4) from the two or more channels (CI; C2; C3; C4) a first channel for that processing of data shall be denied, - memorizing (S4) the first channel, - detecting expiring (S6) of a particular one of first time intervals previously assigned to the first channel, - deciding (S7) that the processing of data from the first channel shall be denied, and - selecting (S8) a further of the two or more channels (CI; C2; C3; C4) for that processing of data shall be denied.

2. Method according to claim 1 comprising the steps of: assigning (S2) to each of the two or more channels (CI; C2; C3; C4) second time intervals outside of which processing of data from a respective channel shall be denied independently from the selection of the first or the further channel, said second time intervals enclosing respective first time intervals, detecting expiring of a particular one of said second time intervals, and deciding that processing of data from a channel previously assigned to the particular one of said second time intervals shall be denied.

3. Method according to claim 1 or 2, wherein a bit rate of at least one of the two or more channels (CI; C2; C3; C4) is time - variant.

4. Method according to any of the preceding claims, wherein the two or more channels are speech channels.

5. Method according to claim 4, wherein the processing of data comprises an encoding or a decoding of speech data.

6. Method according to any of the preceding claims, wherein the processing of data from the two or more channels (CI; C2; C3; C4) comprises a transmitting of data from the two or more transmission channels via a common transmission channel.

7. Method according to any of the preceding claims, wherein the processing of data comprises a processing of data portions and wherein denying processing of data from a determined channel comprises discarding a data portion received on the determined channel.

8. Control unit (CU) for overload handling at a processing device for processing data from two or more channels (CI; C2; C3; C4), comprising a first time table (ST) assigning to each of the two or more channels (CI; C2; C3; C4) first time intervals outside of which processing of data from a respective channel may be selectively denied, a selection unit (SU) for selecting from the two or more channels (CI; C2; C3; C4) a first channel for that processing of data shall be denied, a memory unit (MU) for memorizing the first channel, a timing unit (TU) for detecting expiring of a particular one of first time intervals previously assigned to the first channel (CI), and a decision logic (DL) for deciding that the processing of data from the first channel (CI) shall be denied, wherein the selection unit (SU) is further adapted to select a further of the two or more channels (CI ; C2; C3; C4) for that the processing of data shall be denied.

9. Control unit (CU) according to claim 8 comprising: a second time table (ST) defining for each of the two or more channels (CI ; C2; C3; C4) second time intervals outside of which processing of data shall be denied independently from the selection of the first or the further channel, said second time intervals enclosing respective first time intervals, a timing unit (TU) for detecting expiring of a particular one of said second time intervals, and a decision logic (DL) for deciding that the utilization of the processing of data from a channel previously assigned to the particular one of said second time intervals (CI; C2; C3; C4) shall be denied.