WO2011005158A1

WO2011005158A1 - Method and node for performing voice packet classification in a cellular network

Info

Publication number: WO2011005158A1
Application number: PCT/SE2009/050889
Authority: WO
Inventors: Karl Hellwig; Andreas BERGSTRÖM; Paul Schliwa-Bertling
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2009-07-10
Filing date: 2009-07-10
Publication date: 2011-01-13

Abstract

A method (200) for a node (105, 120; 200) for a cellular system (100), comprising making packet data transmissions to at least one other node (110, 120) and letting said packet data transmissions comprise at least two groups of data packets, speech packets and silence packets. The method comprises analyzing (205) data packets in order to determine a threshold value between said groups of data packets, and classifying (210) data packets as either speech packets or silence packets by means of the threshold value, and classifying (215) data packets of at least one of said groups into one of a number of sub-groups within its group. The method also comprises using (220) the sub-group of a data packet in order to determine an individual priority for a data packet, and using (225) said priority of a data packet when scheduling transmissions of the data packet.

Description

A NODE IN A CELLULAR NETWORK

TECHNICAL FIELD

The present invention discloses an improved node for a cellular telecommunications system, arranged for packet data transmissions to at least one other node in the system. A method for use in such a node is also disclosed. BACKGROUND

VoIP₁ Voice over IP, is one of the service types within cellular systems such as GSM/GERAN, LTE/E-UTRAN and WCDMA which uses transmission of packet data. Within VoIP, there can be different kinds of data packets, such as, for example, speech packets and so called "silence packets", i.e. speech packets which are empty of speech.

As will be realized, different kinds of packets will have different levels of importance for a listener's subjective perception of quality. For example, a speech packet at the beginning of a "talk spurt" will be more important than a speech packet in the middle of a talk spurt, and "silence packets" which mark the beginning of a silent period between talk spurts will be of more importance than silence packets that are intermediate in a silent period. These differences in importance can be used in order to prioritize packets in the event of a "link overload", so as to minimize a perceived quality degradation, particularly so in the example of speech transmission, since a user's speech recognition or understanding is usually such that not all packets need even reach the user. However, in practice, such prioritizing will be difficult to perform for a node in a cellular system such as the GSM or LTE systems, since the node has no immediate way of knowing the contents of a frame. US Patent 7,177,304 B1 , discloses a device which analyzes the voice signals of at least some data speech frames in order to classify each speech frame into one of a plurality of different types of speech. The device also determines a comparative discardability for some of the data speech frames relative to other types of speech, and adds a comparative discardability code, "CDC", to the speech packets.

US Patent Application 2003/0101049 A1 discloses a method and a device by means of which control data signaling may be improved upon with reduced impact on speech quality. According to the disclosure of this document, speech data frames are "stolen" for the purpose of transmitting control data signaling. The disclosure comprises classifying the relative subjective speech signal information content of speech data frames, attaching the classification information to the corresponding speech data frame, and then "stealing" the speech data frames in accordance with the content classification.

SUMMARY

Thus, as has emerged from the description above, there is a need for a solution by means of which a node in a cellular telecommunications system can prioritize between frames in packet data transmissions.

Such a solution is provided the present invention in that it discloses a node for a cellular telecommunications system which is arranged for packet data transmissions to at least one other node in the system. In the packet data transmissions for which the node of the invention is arranged, there is comprised at least two groups of data packets, speech packets and silence packets which are empty of speech.

The node of the invention is arranged to:

- analyze data packets which are to be transmitted in order to determine a threshold value between the two groups of data packets,

- classify data packets which are to be transmitted as either speech packets or silence packets by means of the threshold value,

- classify data packets of at least one of the two groups into one of a number of sub-groups within its group,

- use the sub-group of a data packet in order to determine an individual priority for a data packet,

- use the priority of a data packet when scheduling transmissions of the data packet.

In one embodiment, the node of the invention is arranged to determine the threshold value as a packet size, so that data packets of one of the two groups is smaller than the threshold value and data packets of the other of the two groups is larger than the threshold value.

In one embodiment, the node of the invention is arranged to determine the threshold value as an inter-arrival time, i.e. a frequency of occurrence, so that data packets of one of said groups occur more frequently than the threshold value and data packets of the other of said groups occur less frequently than the threshold value.

In one embodiment, the node of the invention is a BSC, Base Station

Controller, a BTS, Base Transceiver Station, or a User Terminal, a UE, in a

GSM/GERAN system. In other embodiments, the node of the invention is an eNodeB or a UE in an LTE/E-UTRAN system. The invention can also be applied in a WCDMA system, in which case the node of the invention is a UE, a NodeB or an RNC. Thus, as can be seen from these examples, the invention is applicable both for uplink and downlink traffic. In one embodiment, the speech group is the SPEECH packet and the silent group is the SID packet

In one embodiment, the sub-groups of the speech group comprise SPEECH_BEGIN, SPEECH_MID and SPEECH_END.

In one embodiment, the sub-groups of the silent group comprise SID_FIRST, SIDJJPDATE and SID_END.

The invention also discloses a corresponding method for use in a cellular telecommunications system.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will be described in more detail in the following, with reference to the appended drawings, in which

Fig 1 is a schematic overview of a system in which the invention is applied, and

Figs 2-6 are schematic flow charts of different parts of a method of the invention, and

Fig 7 is a block diagram of a node of the invention.

DETAILED DESCRIPTION

The invention will be described in detail in the following with the aid of terminology from the GSM/GERAN standard. It should be emphasized facilitate the reader's understanding of the invention, and should not be used to restrict the scope granted to nor sought for the invention, since the invention can be used in a wide variety of other cellular systems as well, such as, for example, the LTE/E-UTRAN system. Fig 1 shows an overview of a system 100 in which the invention can be applied. The system 100 comprises a number of cells, one of which is shown by way of example as 115 in fig 1. Each cell can accommodate a number of user terminals, UEs, one of which is shown by way of example as 120 in fig 1.

For each cell in the system 100, there is also a controlling node, through which all traffic to and from the UEs in the cell is routed. In GSM, the controlling node of a cell is usually referred to as the BTS, Base Transceiver Station. The BTS of the cell 115 is shown as 110 in fig 1. Apart from the UE and the BTS, there are also "higher" nodes in the system 100, one of which is the so called BSC, the Base Station Controller, which is shown as 105 in fig 1. As implied by the name, one function of the BSC is to control one or more BTS in the system 100. The traffic between the UE 120 and the BTS 110 can be either of the circuit switched, CS, or packet switched, PS. The present invention deals with the case of packet switched traffic, one example of which is VoIP, Voice over IP, and aims at giving the BTS the possibility of prioritizing between packets in a better way than previously. However, the invention can also be applied in a BSC, since traffic to and from the BTS will usually go via the BSC.

In packet switched application such as VoIP, the traffic to/from the BSS and the BTS is in the form of so called LLC PDUs, Logical Link Control Packet Data Units. Thus, it is these LLC PDUs which will be prioritized by means of the present invention. Fig 2 shows a flow chart of a basic embodiment of a method 200 of the invention. Steps which are options or alternatives have been indicated by means of dashed lines in fig 2.

As shown in step 202 of fig 2, the method is performed on an incoming LLC PDU, here shown as "PDU X" in order to signify that the LLC PDU is one in a series of PDUs, i.e. PDU 1 , PDU 2 Suitably but not necessarily, the method is performed on consecutive PDUs.

As shown in step 205, the method 200 comprises analyzing an incoming LLC PDU in order to obtain a threshold value, "T", between two groups of data packets, speech packets and silence packets which are packets that are empty of speech. The threshold value T is used in order to classify, step 210, the incoming LLC PDU into one of two groups, Speech or "SID", where SID is a silent packet, i.e. one which contains no speech.

As shown in step 215, the method 200 also comprises classifying the LLC PDUs into "sub-groups", which are then used, step 220, in order to prioritize between different LLC PDUs. As shown in step 225, the priority is used when scheduling onward transmissions of the LLC PDUs.

As indicated in step 216, the method of the invention also in one embodiment comprises remembering the classification of a PDU, and using that classification in order to perform the classification of a subsequent PDU. In step 216 this is shown as remembering the classification of PDU X-1 in order to classify PDU X. How a remembered PDU classification can be used will be elaborated upon later in this text. Fig 3 shows a more detailed flow chart of steps 205 and 210 of fig 2, i.e. how an incoming PDU is analyzed in order to arrive at a threshold value between the two PDU groups SPEECH and SID. The threshold used in this example will be a size threshold, since, in one kind of coding, so called AMR, Adaptive Multi Rate coding, speech frames will always be larger than SID frames. However, this is merely an example of a possible threshold, since, in other kinds of speech coding, this is not necessarily true. In such cases, the fact that SID frames occur more isolated, whereas the speech frames occur more frequently, may instead be used in order to arrive at a threshold. Thus, the size threshold may be replaced by inter-arrival-time thresholds, but otherwise the logic and mechanisms used would be the same. Of course, other alternative threshold categories are also be possible.

Thus, the method which will be described in the following, with reference to fig 3 will be one which is performed adaptively based on statistics. When receiving LLC PDUs, the method is used to collect statistics and to thus determine the size of the SID frames and speech frames respectively by the sizes received and the sequence in which they occur. This is partly made possible by the fact that SID frames are typically followed by a speech pause and are then typically repeated sporadically to update the so called "Comfort Noise" and to show that the call is still ongoing. Hence, SID Frames occur isolated most of the time, as opposed to which, speech frames typically occur in series and are never isolated, unless of course, a transmission error has occurred.

In order for the proposed classification method to occur, it is valuable if the so called "QoS", Quality of Service, is used and that the data flow in which the LLC PDUs arrive has the QoS class "Conversational", and that the Source Statistic Descriptor, SSD, is set to "Speech". However, the invention can also be used if, for example, the QoS Class is "Streaming", although the expected gains of the invention may be lower. Thus, a brief description of this, with reference to fig 3 is that the method 300 is used for an incoming PDU, step 302, and a check is made in step 305 to see if the QoS class is conversational. If this is not the case, the method waits for the next PDU, i.e. goes back to step 302. If however, the QoS class is conversational, the method goes onto step 315.

In step 315, a check is made of whether or not the SSD is set to speech, and if this is not the case, the method waits for the next PDU, i.e. goes back to step 302. If the SSD is set to speech, a size threshold T is determined in step 320, following which the size of the PDU is compared to the threshold. If the PDU size exceeds the threshold T, the PDU is classified as a SPEECH PDU, step 325, and if not, the PDU is classified as a SID PDU, step 330. Fig 4 shows a flow chart of an example of a method 400 for how step 317," set T", of fig 3 may be carried out, i.e. how the threshold T is obtained. As shown in step 402, the method is performed for incoming LLC PDU X. Two variables are used in this method, MIN_SIZE and MAX_SIZE. Initially, MIN-SIZE should be set to "positive infinity", i.e. "+•»", and MAX_SIZE should be set to zero, i.e. "0".

As shown in step 402, a check is made to see if the size of the incoming PDU is below the size of the variable MIN-SIZE. and the method proceeds to step 412If this is the case, the variable MIN_SIZE is changed to the size of the incoming LLC PDU, step 406. If this is not the case, the size of the incoming PDU is compared to the size of the variable MAX_SIZE, step 408.

If the incoming PDU is larger than the variable MAX_SIZE, then the variable MAX_SIZE is set to the size of the incoming PDU, step 410, and the method proceeds to step 412. In step 412, the size of the variable MIN_SIZE is compared to the size of the variable MAX_SIZE. If MIN_SIZE is the smaller of the two, the threshold T is set to the size of the variable MIN_SIZE, step 414, and if not, the incoming PDU is defined as a SPEECH PDU, step 416.

So far, we have seen how the threshold T may be set, and how the classification of the incoming LLC PDUs can be carried out according to the invention. In the following, a description will be given of how the incoming PDUs can be classified into the mentioned "sub-groups", e.g. for the case of SPEECH PDUs the sub-groups "SPEECH_BEGIN", "SPEECH_MID" and "SPEECH_END", and for the SID PDUs the sub-groups "SID FIRST", "SID UPDATE" and "SID END". The invention uses the fact that by taking into account the SID or SPEECH classification of the currently processed LLC PDU, i.e. PDU X, and also the previous LLC PDU, i.e. PDU X-1 , it is possible to detect the beginning and also the end of a "talk spurt": If the beginning of a talk spurt is detected, the current LLC PDU X, as well as the following ones, are classified as 'SPEECH_BEGIN\ How many these will be are determined explicitly by keeping track of the number of received LLC PDUs until a predefined parameter value is reached. All previous LLC PDUs, if still queued for transmission but not yet transmitted, are classified as 'SID_END'. Thereafter, all new LLC PDUs are classified as 'SPEECH_MID'.

If, on the other hand, the end of a talk spurt is detected, e.g. by detection of a following SID frame, then this LLC PDU is classified as 'SID-FIRST', following which any queued LLC PDUs are re-classified as 'SPEECH_END'. Finally, in the case that one SID frame is detected after another SID frame, then the latter SID frame will be classified as SID UPDATE. The length of the beginning of a talk spurt and the length of the end part of a talk spurt have both been assumed to be configurable parameters, which can be varied from system to system or from installation to installation. As an alternative solution, instead of letting these values represent an explicit number of LLC PDUs and perform counting as described above, an alternative according to the invention is to let these values represent time intervals. By means of the invention as described so far, a threshold T has been determined, and incoming PDUs have been classified into both one of the "major groups" SPEECH or SID, and into one of their sub groups, i.e. SPEECH_BEGIN', 'SPEECH_MID', 'SPEECH_END', ¹SID-FIRST¹, 'SIDJJPDATE' or ¹SID-END'. The classification of the incoming PDUs is also used by the invention in order to determine the individual priority of each incoming PDU, in a manner which will be described in the following with reference to the flow chart of fig 5 which schematically shows a method 500 for this. As shown in fig 5, the classification of the PDU, i.e. "PDU X", is used in order to determine the PDU's priority as follows: initially, step 505, the classification of an incoming LLC PDU, i.e. PDU X is determined. The classification is used as follows: • PDU classification SPEECH END or SID END: Low priority, step 520.

• PDU classification SPEECH BEGIN or SID FIRST: High priority, step 530.

• PDU classification SIDJJPDATE: Check, step 510, if the previous PDU was discarded by the RRM, the Radio Resource Management, and if this was the case, then give medium priority to the PDU, step 525, otherwise Low priority, step 520.

• PDU classification SPEECH_MID: Check, step 515, if the previous PDU was discarded by the RRM, the Radio Resource Management, if not, then give Medium priority to the PDU, step 525, otherwise give high priority, step 525.

A next step according to the invention is to use the priority information as described above when scheduling the transmission of the PDU, for example by an RRM functionality. Some illustrative examples of how this may be done will be given in the following.

Fig 6 shows a flow chart of a method 600 which shows how the priority information of a PDU as obtained by means of the invention can be used in order to optimize scheduling: Given the individual priorities of the LLC PDUs, high, medium or low, which is determined in step 605, an RRM scheduler can prioritize the scheduling via the radio resources according to the priorities as shown in fig 6, i.e. a high priority PDU is given high transmission priority, step 610, and similarly for medium and low priority PDUs, steps 615 and 620 respectively in the scheduling, step 625. This may be quite beneficial e.g. in cases of, where the network is not able to fit all PDUs on the available resources. A further example of how the priority classification as obtained by means of the invention can be used is as follows: within GSM/GERAN, the possibility exists to let the Radio Link Control, RLC, entities work in so called Non- Persistent, NPM, mode. In this mode, RLC packets may be retransmitted just as for RLC Acknowledged mode, with the difference that the number of retransmissions is limited in order to keep the delay jitter limited. The number of retransmissions that is allowed for each RLC PDU is implicitly determined by the RLC NPM timer. Thus, by adjusting the value of this timer on a per- packet basis, it would be possible to allow higher-priority LLC PDUs more retransmissions than medium- or low-priority ones

A second further example of how the priority classification as obtained by means of the invention can be used is as follows:

A link adaptation module selects the appropriate modulation and abound of channel coding depending on the radio environment as well as other factors such as BLER targets, service profiles etc. It could therefore be an option to let the link adaptation module choose a more robust combination of modulation and coding (called Modulation and Coding Scheme, MCS, in

GERAN) if the LLC PDU is deemed to have a high priority and a less robust such MCS if the LLC PDU is of lesser importance etc

A third further example of how the priority classification as obtained by means of the invention can be used is to let a power control module command the usage of different transmission power levels depending on the importance of the PDU which is to be transmitted.

Fig 7 shows an example of a block diagram 700 of a generic node of the invention, which can be one of the following: In a GSM/GERAN system:

• a BSC, Base Station Controller, or

• a BTS, Base Transceiver Station, or

• a User Terminal, a UE. In an LTE/E-UTRAN system: • an eNodeB or

• a UE.

In a WCDMA system:

• an RNC, or

• a NodeB or

• a UE.

As shown in fig 7, the generic node 700 comprises an interface 705 towards other units or nodes in the system. The interface 705 can, for example, be an antenna interface or an interface to a "landline", so that the node 700 can communicate with other units or nodes in the system.

The node 700 also comprises a receive unit 710 and a transmit unit 715. The interface 705 is suitably an interface for transmission and a reception, so that the receive and transmit units 710, 715, can both be connected to the antenna unit.

The receive and transmit units 710, 715, are controlled by a processing unit 720, which can also be seen as a control unit, and the processing unit is connected to a memory unit 725 which is used for, for example, storage of information and, where applicable, executable program code. In fig 7, the processing unit 720 is shown as "μP", in order to indicate that this unit suitably comprises one or more microprocessors. This should naturally only be seen as an example, other kinds of processing units or control units are also perfectly possible within the scope of the present invention.

The interface unit 705, the transmit unit 715, the receive unit 710, the processing unit 720 and the memory unit 725 enable the controlling node 700 to communicate with and handle the control of one or more user terminals in a cell such as the cell 125 in the system 100. The processing unit 720, together with the memory unit 725, performs the following tasks:

- analysis of data packets which are to be transmitted in order to determine a threshold value between said groups of data packets,

- classification of data packets, which are to be transmitted as either speech packets or silence packets by means of said threshold value,

- classification of data packets of at least one of said groups into one of a number of sub-groups within its group,

- using the sub-group of a data packet in order to determine an individual priority for a data packet,

- using the priority of a data packet when scheduling transmissions of the data packet.

In addition, the processing unit 720, together with the memory unit 725, also, when applicable, determines the threshold value as a packet size, so that data packets of one of said groups is smaller than the threshold value and data packets of the other of said groups is larger than the threshold value.

Alternatively, when applicable, the processing unit 720, together with the memory unit 725, determines the threshold value as an inter-arrival time, i.e. a frequency of occurrence, so that data packets of one of said groups occur more frequently than the threshold value and data packets of the other of said groups occur less frequently than the threshold value. Some other tasks which, where applicable, are performed by the processing unit 720, together with the memory unit 725 are:

- detecting if the GSM QoS, Quality of Service, is used, and if not, interrupting the analysis of the data packets, - detecting if the QoS class is "Conversational", and if this is the case, the analysis proceeds and if not, the analysis of data packets is interrupted.

- detecting if the GSM Source Static Descriptor, the GSM SSD, is set to "Speech", and if this is the case, the analysis proceeds, and if not, the analysis of data packets is interrupted.

The invention is not limited to the examples of embodiments described above and shown in the drawings, but may be freely varied within the scope of the appended claims.

Claims

1. A node (105, 120; 200) for a cellular telecommunications system (100), the node being arranged for packet data transmissions to at least one other node (110, 120) in the system, in which packet data transmissions there are comprised at least two groups of data packets, speech packets and silence packets which are empty of speech, the node (105, 120) being characterized in that it is arranged to:

- analyze (205) data packets which are to be transmitted in order to determine a threshold value between said groups of data packets,

- classify (210) data packets, which are to be transmitted as either speech packets or silence packets by means of said threshold value,

- classify (215) data packets of at least one of said groups into one of a number of sub-groups within its group,

- use (220) the sub-group of a data packet in order to determine an individual priority for a data packet,

- use (225) said priority of a data packet when scheduling transmissions of the data packet.

2. The node (105, 120; 200) of claim 1 , being arranged to determine said threshold value as a packet size, so that data packets of one of said groups is smaller than the threshold value and data packets of the other of said groups is larger than the threshold value.

3. The node (105, 120; 200) of claim 1 , being arranged to determine said threshold value as an inter-arrival time, i.e. a frequency of occurrence, so that data packets of one of said groups occur more frequently than the threshold value and data packets of the other of said groups occur less frequently than the threshold value.

4. The node of any of claims 1-3, being one of the following in a GSM/GERAN system:

• a BSC, Base Station Controller, or

• a BTS, Base Transceiver Station, or

• a User Terminal, a UE.

5. The node (105, 120; 200) of claim 4 or 5, in which node the speech group is the SPEECH packet and the silent group is the SID packet

6. The node (105, 120; 200) of claim 5, in which said sub-groups of the speech group comprise SPEECH_BEGIN, SPEECH_MID and SPEECH_END.

7. The node (105, 120; 200) of claim 5 or 6, in which said sub-groups of the silent group comprise SID_FIRST, SIDJJPDATE and SID_END.

8. The node (105, 120; 200) of any of claims 4-7, being arranged to detect if the GSM QoS, Quality of Service, is used, and if not, the analysis of data packets is interrupted.

9. The node (105, 120; 200) of claim 8, being arranged to detect if the QoS class is "Conversational", and if this is the case, the analysis proceeds and if not, the analysis of data packets is interrupted.

10. The node (105, 120; 200) of claim 8 or 9, being arranged to detect if the GSM Source Static Descriptor, the GSM SSD, is set to "Speech", and if this is the case, the analysis proceeds, and if not, the analysis of data packets is interrupted.

11. The node of any claims 1-3, being an eNodeB or a UE in an LTE/E- UTRAN system.

12. The node of any claims 1-3, being an RNC, a NodeB or a UE in a WCDMA system.

13. A method (200) for use in a node (105, 120; 200) for a cellular telecommunications system (100), the method comprising making packet data transmissions from the to at least one other node (110, 120) in the system, and letting said packet data transmissions comprise at least two groups of data packets, speech packets and silence packets which are empty of speech, the method being characterized in that it comprises:

- analyzing (205) data packets which are to be transmitted in order to determine a threshold value between said groups of data packets,

- classifying (210) data packets, which are to be transmitted as either speech packets or silence packets by means of said threshold value,

- classifying (215) data packets of at least one of said groups into one of a number of sub-groups within its group,

- using (220) the sub-group of a data packet in order to determine an individual priority for a data packet,

- using (225) said priority of a data packet when scheduling transmissions of the data packet.

14. The method (200) of claim 13, further comprising determining said threshold value as a packet size, so that data packets of one of said groups is smaller than the threshold value and data packets of the other of said groups is larger than the threshold value.

15. The method (200) of claim 13, further comprising determining said threshold value as an inter-arrival time, i.e. a frequency of occurrence, so that data packets of one of said groups occur more frequently than the threshold value and data packets of the other of said groups occur less frequently than the threshold value.

16. The method (200, 210) of any of claims 13-15, according to which the speech group is the SPEECH FRAME and the silent group is the SID FRAME.

17. The method (200) of claim 16, applied to a BSC, Base Station Controller, or a BTS, Base Transceiver Station, in a GSM system.

18. The method (200) of claim 16, applied to an eNodeB or a UE in an LTE system.

19. The method (200) of claim 16, applied to a UE, a NodeB or an RNC in a WCDMA system.

20. The method (200) of any of claims 16-19, in which said sub-groups of the speech group comprise SPEECH_BEGIN, SPEECH_MID and SPEECH_END.

21. The method (200) of any of claims 16-20, in which said sub-groups of the silent group comprise SID_FIRST, SiDJJPDATE and SID_END.

22. The method (200) of any of claims 16-21 , comprising detection of whether or not the GSM QoS, Quality of Service, is used, and if interrupting the analysis of data packets if this is not the case.

23. The method (200) of claim 22, comprising detection of whether or not the QoS class is "Conversational", and if this is the case, proceeding with the analysis, and if not, interrupting the analysis of data packets.

24. The method (200) of claim 22 or 23, comprising detection of whether or not the GSM Source Static Descriptor, the GSM SSD, is set to "Speech", and if this is the case, proceeding with the analysis, and if not, interrupting the analysis of data packets.