FR2792787A1 - Balancing the codependent ratio of Eb/l in a service multiplexing code division multiple access system in a telecommunication system - Google Patents

Abstract

Each transport channel (100) periodically receives a transport block set from a higher level (102) with a number of blocks depending on the size of the blocks and on the transport channel. Transport channels with the same quality of service are processed by the same processing channels (103A,B,C) and, after channel encoding and interleaving, the blocks are multiplexed together by concatenation (104). Each processing process includes a rate matching step (112,114) to convert the block size according to the Eb/l ratio and the maximum puncture rate corresponding to the quality of service. Independent claims are included for a base station and a mobile station.

Description

<Desc / Clms Page number 1>

BACKGROUND OF THE INVENTION
The 3GPP group is an association whose members come from several regional standardization bodies including ETSI and ARIB, and whose goal is the standardization of a third generation telecommunication system for mobiles. One of the fundamental aspects distinguishing third generation systems from second generation systems is that, in addition to using the radio spectrum more efficiently, they will allow a very high level of service flexibility. Second generation systems offer an optimized radio interface for certain services, for example GSM is optimized for speech transmission (telephony). Third-generation systems will provide a radio interface for all kinds of services and service combinations.

au moins un certain Ebll minimum dépendant du codage. Le Eb/1 exprime le rapport entre l'énergie de chaque bit codé, et la puissance des interférences. One of the challenges of third-generation mobile radio systems is therefore to efficiently multiplex on the radio interface services that do not have the same requirements in terms of quality of service (QoS). In particular, these differences in quality of service imply respective transport channels having different codings and channel interleaves, and requiring different maximum bit error rates (BER). For a given channel coding, the requirement on the BER is fulfilled when the coded bits have

at least some minimum Ebll depending on the coding. The Eb / 1 expresses the ratio between the energy of each coded bit, and the power of the interferences.

It follows that the different qualities of service do not have the same requirement in terms of Eb / l. However, in a CDMA-type system, the capacity of the system is limited by the level of interference, it is therefore necessary to set the Eb / l to the most just for each service. It follows that a flow balancing operation in order to balance the Eb / l is required between the different services. Without this operation the Eb / l would be set by the service with the highest requirement, and it would result that the other services would be too good quality which would directly impact the capacity of the system.

<Desc / Clms Page number 2>

 This is problematic because it is necessary in a certain way that the balancing reports of flow are defined identically to both ends of the radio link.

 The present invention relates to a method for defining equalization rate balances identically to both ends of a CDMA type radio link.

PRIOR ART
In the ISO OSI model a telecommunication equipment is modeled by a layered model constituting a protocol stack where each level is a protocol providing a service at the top level.

The service provided by level 1 is called transport channels. A transport channel allows the higher level to transmit data with a certain quality of service. The quality of service is particularly characterized by the delay and the BER (English acronym designating the bit error rate). In order to fulfill the quality of service requirement, Level 1 uses some suitable channel coding and interleaving.

 FIGS. 1 and 2 show the interleaving and multiplexing block diagrams as defined by the present proposal of the 3GPP group, this proposal not being yet fixed or completely clear. In these figures similar blocks bear the same numbers. The upstream link (from the mobile station to the network) is distinguished from the downlink (from the network to the mobile station), and only the transmission part is represented.

 For each transport channel the higher levels provide (122A and 122B) at level 1 (L1) periodically a set of transport blocks. The number of transport blocks in this set, as well as their sizes, depend on the transport channel. The minimum period with which the transport block set is provided is called the transport channel transmission interval and corresponds to the time span of its channel interleaving. The transport channels, after channel coding are balanced in flow, then interleaved and multiplexed with each other. This multiplexing is performed by multiplexing frame: multiplexing frame is the smallest unit of data for which the demultiplexing can be

<Desc / Clms Page number 3>

 operated at least partially. A multiplexing frame typically corresponds to a radio frame. The radio frames form consecutive time intervals synchronized on the network, and numbered by the network.

In the proposal of 3GPP group a radio frame corresponds to a duration of 10ms.

 The 3GPP proposal includes the option of service-specific coding and interleaving (122B, 118). The possibility of such an option is considered to date because it has not been determined whether it is essential. Indeed, such an option is not welcome, and the manufacturers of mobile telecommunications networks want to have a level 1 as generic as possible in order to reuse the same calculation means for a wide variety of services.

 In the general case (130) a FCS (Frame Check Sequence) is affixed (100) to each transport block. The FCS is typically calculated by the so-called CRC (Cyclic Redundancy Check) technique which consists of considering the bits of the transport block as the coefficients of a polynomial P and calculating the CRC from the remainder of the polynomial (P + PO) in the division by a generator G polynomial, where PO is a predefined polynomial for a given degree of P. Applying the FCS is optional, and some transportation channels do not include this step. The exact calculation technique of the FCS also depends on the transport channel, and in particular the maximum size of the transport blocks. The purpose of the FCS is to detect whether the received transport block is valid or corrupt. This makes it possible to implement antenna selection technique in the event of diversity of reception sites in the network, as well as automatic repetition protocols.

 The next step (102) is to multiplex together the transport channels (TrCH) of the same quality of service (QoS). Indeed, these transport channels having the same quality of service, they can use the same channel coding. Typically the multiplexing is operated by concatenating the sets of transport blocks with their FCS for each transport channel.

 The next step is to perform channel coding (104). At the output of the channel encoder is a set of coded blocks. Typical-

<Desc / Clms Page number 4>

 This set contains zero or an encoded block (the coded zero block case occurring when zero transport blocks have been supplied to L1).

 It is from this stage that the rising link is distinguished from the descending link.

 The flow balancing step (124, 108) is intended to balance the Eb / l ratio between the different service quality transport channels. The ratio Eb / l gives the average energy of a bit compared to interference, in a system using multiple access CDMA technology, the higher the ratio, the higher the quality that can be achieved. It is thus understood that transport channels having different qualities of service do not have the same need in terms of Eb / l, and that in the absence of flow balancing, certain transport channels would be "too much" quality. good because fixed by the most demanding channel, and they would unnecessarily cause interference. Flow balancing therefore has a balancing role of Eb / l. The flow balancing is such that X bits at the input give Y bits at the output, which multiplies Eb / l by the ratio Y / X, hence the possibility of balancing.

effet la somme Yi+Y2+...Y|< des nombres de bits après équilibrage doit être égale ou nombre total de bits dans la trame radio pour les données. Ce nombre ne peut prendre que certaines valeurs prédéfinies N1, N2, ...,Nr. Il convient donc de résoudre le système à k inconnues Y1, ..., Yk suivant : Flow balancing is not done in the same way in the upstream link and in the downlink. Indeed, in the uplink it was decided to transmit continuously, because a discontinuous transmission deteriorates the peak / average ratio of the radiofrequency power output of the mobile station. The closer this ratio is to 1, the better. Indeed, if this ratio is deteriorated (= increased) this means that the power amplifier requires a larger margin (backoff) of linearity with respect to the average point of operation. Due to such a margin, the power amplifier would be less efficient, and therefore consume more for the same average power emitted, which would reduce unacceptably the battery life of the mobile station. Since it is necessary to transmit continuously on the rising link the ratio Y / X ratio can not be constant. In

effect the sum Yi + Y2 + ... Y | <of the numbers of bits after balancing must be equal or total number of bits in the radio frame for the data. This number can only take certain predefined values N1, N2, ..., Nr. It is therefore necessary to solve the system with k unknowns Y1, ..., Yk according to:

<Desc / Clms Page number 5>

où X, et Eb@/l et P, sont des constantes caractéristiques de chaque canal de transport, et où on cherche à minimiser Nj parmi les p valeurs possible N1, N2, ...,Nr (note : P, est le taux de poinçonnage maximal supportable par un canal de transport codé).

where X, and Eb @ / l and P, are characteristic constants of each transport channel, and where one seeks to minimize Nj among the possible p values N1, N2, ..., Nr (note: P, is the rate maximum punching tolerable by a coded transport channel).

 Thus in the link up the Y / X rate balancing ratios for each transport channel are not constant from one multiplexing frame to the next, but they are defined at a multiplicative constant: the two-to-two ratios between these reports therefore remain constant.

 In the downlink, the peak / average ratio of radio frequency power is in any case very bad because the network transmits to several users simultaneously, and the signals intended for these users combine in a constructive or destructive manner, thus inducing wide variations of power. radiofrequency emitted by the network, and therefore a poor peak / average ratio. It was therefore decided that for the downlink the Eb / l balancing between the different transport channels would be done with constant Y / X ratio flow balancing, and that the multiplexing frames would be supplemented with dummy bits, c ie non-transmitted bits, ie a discontinuous transmission.

 Thus, the difference between the upstream link and the downlink is that in the up link the rate balancing 108 is dynamic so as to complete the multiplexing frames, whereas in the downlink the rate balancing 124 is static, and the Completion of the multiplexing frames is done by inserting dummy bits 206.

SMG2/UMTS-L 1 /Tdoc428/98. Cet algorithme permet d'obtenir des rapports de poinçonnage/répétition qui ne sont pas entiers, et il est donné pour information dans la table 1. Cet algorithme a été depuis la proposition de Sie- Flow balancing, whether dynamic or static, is done either by repetition or by bit punching, according to an algorithm that has been proposed to ETSI by Siemens in the referenced technical document.

SMG2 / UMTS-L 1 / Tdoc428 / 98. This algorithm makes it possible to obtain punching / repetition ratios that are not integers, and it is given for information in table 1. This algorithm has since been proposed by Sie

<Desc / Clms Page number 6>

 modified for the uplink link (mobile station to network) so as to better cooperate with the interleaver 130 which precedes it.

 The peculiarity of this algorithm is that when it operates in punching mode it avoids punching consecutive bits, but on the contrary tends to space a maximum of two punched bits. For repetition, the repeat bits follow the bits they repeat. Under these conditions it is understood that it is advantageous that the flow balancing is done before interleaving. Indeed for the repetition this allows to discard the repeated bits by the fact that interleaving follows the flow balancing. For punching, the fact that an interleaver precedes the rate balancing runs the risk that the rate balancer punctures consecutive bits at the output of the channel encoder.

 It is therefore advantageous that the flow balancing be made as high as possible, that is to say as close as possible to the channel coder.

 For the downlink it is possible to set the flow balancing just at the output of the channel coding 104, because the flow balancing report is constant. Therefore, only one interleaver (134) is needed, however a second interleaver (136) is required because the multiplexing of the different QoS transport channels is by simple concatenation, and such a process would effectively limit the time span of each multiplexed block.

 For the link up the rate balancing report varies at all the multiplexing frames, therefore, at least one first interleaver (130) is required before the rate balancing to allocate the bits of the coded block on the plurality of multiplexing frames, and a second interleaver (132), placed after the rate balancing to separate between them the bits repeated by the rate balancing. Also, a step 120 of multiplexing frame segmentation is between the 1st and the 2nd interleaver. This step consists in segmenting the coded and interleaved blocks by the first interleaver in as many segments as is the ratio between the temporal extent of the first interleaver and the duration of a multiplexing frame.

This segmentation is typically done so that the concatenation of the segments restores the interleaved coded block.

<Desc / Clms Page number 7>

 Note that in the upstream link this step 120 of segmentation is necessarily before the flow balancing 108. Indeed the flow balancing 108 is done according to a dynamically established ratio multiplexing frame multiplexing frame, it is not therefore not possible to do it on a unit of data that can extend over several multiplexing frame.

 Currently only the multiplexing, channel coding, interleaving, and rate balancing algorithms are defined and discussed. There is no rule for setting how a size X of an input block of the balancer is associated with a size Y at the output. We are reduced to assuming that all X # Y pairs are predefined in memory.

PROBLEM OF THE PRIOR ART PROBLEM OF THE SIGNALING OF THE CONNECTIONS IN THE AMOUNT
A rule for determining the Y size of a balanced rate block with the other blocks, starting from the X size of this block before flow balancing is required at least in the uplink link. Indeed, because the services are variable rate, the number of transport blocks provided for each transport channel is variable. The list (X1, X2, ..., Xk) of the block sizes to be balanced in the rate can therefore vary from multiplexing frame to multiplexing frame (the number k of elements in this list is also not necessarily constant).

 Since the size Y, associated with the size X, does not depend only on Xi but on the whole list (X1, X2, ..., Xk) because of the dynamic balancing, it follows that we have a list (Y1, Y2, ..., Yk) for each list (X1, X2, ..., Xk). The number of lists can therefore be very large, at least as large as the number of combinations of transport formats (a combination of transport format is a quantity defining how to demultiplex the multiplexing frame).

 Thus storage in memory of all associations (X1, X2, ..., Xk) # (Y1, Y2, ..., Yk) can be prohibitive. This is all the more true that it would be necessary to have such an association list for each combination

<Desc / Clms Page number 8>

 of services possible. One solution would be to report this association list by the higher levels. But then the cost of signaling to establish a new transport channel would be high.

PROBLEM OF NEW EB / L ADJUSTMENTS
In addition, the exact Eb / l balancing depends on the channel decoder technology for each QoS. The performance of such a device may vary from one manufacturer to another according to their respective know-how.

In fact this balancing does not depend on the absolute performance of each decoder, but their relative performance, which can vary from one manufacturer to another, if the performance of one of them varies.

 Imagine two qualities of service A and B, and two constructors M and N. M and N have the same channel decoder for A, but M has a decoder much more efficient than that of N for B. It is then clear that the constructor M could benefit from a stricter Eb / l for B. Indeed this would decrease the total power needed, and thus would produce a gain in capacity that would allow M to sell more mobile telecommunication equipment to network operators by arguing that that.

 It would therefore be very useful to be able to report parameters allowing to define the rule of determination X # Y of the size Y after flow balancing, this would make it possible to negotiate or renegotiate the proportions of Eb / l.

This signaling must be the least expensive possible.

 This dynamic adjustment of Eb / l done by the higher levels therefore means that if two telecommunication stations A and B want to establish or modify a connection on which there is service multiplexing, then they follow the following steps: 1. B signals At what is the maximum load N of a multiplexing frame that B can transmit.

2. A determines the ideal proportion for A of Eb / l from: - the value of N received from B - the maximum punching rate that A supports for each QoS, - the relative requirements of each QoS in terms of Eb / l - the minimum performance requirement specified for A

<Desc / Clms Page number 9>

 4. A tells B what proportion of Eb / l A is expected.

 Note that step 1 is not necessarily present, one can indeed imagine systems or it is known in advance and be part of the characteristics of the system. That said, such a system would be very unlikely given their lack of flexibility.

 It is possible that the proportion of Eb / l determined by A is suboptimal compared to the goal that no transport channel has more than what it deserves. This is a compromise situation in which it is preferred to reduce the capacity of the network provided that the service combination connection can be established.

 Such a compromise is acceptable to the extent that the degradation is within the limits set by the minimum performance requirement defined in the system specification.

 It is also possible that the actual tolerance limit is partly at the discretion of the network. This would define unsecured service levels, in which the service is provided when traffic conditions permit, and otherwise it is renegotiated downward.

PROBLEM OF DETERMINING COMBINATION ASSEMBLIES OF POSSIBLE TRANSPORT FORMAT.

 There will certainly be a specification of possible combinations of services. In this specification, for each service combination, will be associated a set of combinations of transport formats. This will surely be the case for basic services such as the conventional telephony service, and all the associated services (call signal, call waiting, etc.).

...)
However, the number of potential combinations may well increase in the future, and then clear rules will be needed for higher levels to determine if such combinations are possible, and how to negotiate them, and / or renegotiate them, and also to determine the set of transport format combinations for a given combination.

 It should therefore be possible for the higher levels, using simple arithmetic algorithms, to determine which combinations of

<Desc / Clms Page number 10>

Transport formats are possible. To do this, there are at least three arithmetic rules that higher levels should apply: # The first rule converts the number of elements in the transport block sets and their respective sizes into the number of elements. sets of coded blocks and their respective sizes,
For example, this rule can be of type Y = X / yield + Nqueue, where yield and Nqueue are characteristic constants of the code.

# The second rule converts the size of an encoded block into the size of a segment produced by segmentation 120 per multiplex frame. In general, this rule is a simple division by F when the transmission interval of the corresponding transport channel corresponds to the multiplexing frame. However it is not yet clear if the segmentation is equal, ie the coded blocks have a size that is a multiple of
F, or unequal, ie the size of the segment is set to 1, and it is necessary to know the sequence number of the segment to reduce the ambiguity.

 # The third rule is that which makes it possible to deduce from the size X of a block to balance in flow, the size Y of the block balanced in flow.

 This third rule remains to be specified.

noté El,,E2, Ep, alors les Eb/I de chaque qualité de services seront dans les mêmes proportions que les coefficients Ei. SOLUTION SUPPLIED BY THE INVENTION GENERAL PRINCIPLE
In the invention each quality of service is characterized by two integers E and P. E corresponds to the Ebl1, that is to say that if there are several service qualities 1,2, p, whose coefficients E respective

denoted El ,, E2, Ep, then the Eb / I of each quality of services will be in the same proportions as the coefficients Ei.

 The coefficient P is the maximum punching rate that is acceptable for a given quality of service.

 We use integers because: - Calculations on integers, or fixed-point calculations, are simpler to implement, in other words they are done more quickly or with less resources,

<Desc / Clms Page number 11>

 - The precision of the calculations on the integers is very easily quantifiable by the number bits of the registers in which these integers are stored. Thus one can easily ensure that the same rounding errors are produced in the network and in the mobile station, and therefore the result of the calculations is exactly the same on both sides of the radio interface.

On a PMAX<PBASE, et PBASE est le taux de poinçonnage maximal. DYNAMICS OF PARAMETERS E AND P AND OTHER CONSTANTS
The dynamics is defined as follows: # E is a 1 integer from 1 to EMAX, # P is an integer from 0 to PMAX,

PMAX <PBASE, and PBASE is the maximum punching rate.

 Thus, the algorithm of the invention is characterized by 3 integer constants EMAX, PMAX and PBASE.

 Subsequently we will see a fourth constant LBASE whole which relates to the accuracy of calculations.

 Note that although we use the same EMAX, PMAX, PBASE and LBASE notations for the uplink (mobile-to-network) link and the downlink (network-to-mobile) link, the corresponding constants do not necessarily have the same value in both cases.

X AND Y RATINGS
Subsequently we will also use the same notation X and Y for the up link and the downlink with different meanings.

 In addition we will define a Q application giving the QoS value for a given index.

<Desc / Clms Page number 12>

RATINGS FOR DOWNLINK
In the downlink we write X1, X2, ..., Xk the list of sizes before flow balancing possible for the blocks for a given quality of service (QoS), and this for all quality of service (QoS) values. possible.

To be more precise, if the QoS takes values from 1 to p, then:
Xko + 1, ..., Xk1 are all possible block sizes for 1 QoS
Xk1 + 1, ..., Xk2 are all possible block sizes for 2 QoS Xkp-1 + 1, .... Xkp are all possible block sizes for p QoS with the convention that ko = 0 and kp = k and ko <ki <... <kp.

Moreover, we consider an application Q of the set {1, k} of the block size indexes for a QoS to the set of service quality indexes {1,..., P}. So we have :

Note that given the preceding definitions it is possible to have twice the same block size (X, = Xj with i # j) provided that there is not the same quality of service (Q (i) Q (j)).

RATINGS FOR THE AMOUNT
For the uplink, we number 1,2, ..., k the blocks that are to be balanced in flow for a given multiplexing frame, and X1, X2, ..., Xk are their respective sizes.

 Thus the list (X1, X2, ..., Xk) varies from multiplexing frame to multiplexing frame, and moreover its element number k is also not necessarily constant.

 Q is an application of {1, ..., k} to {1, ..., p}, which for the considered multiplexing frame, associates with index i of a block its quality of service Q (i ).

<Desc / Clms Page number 13>

même taille de bloc (Xi = Xj avec i j) qu'ils aient ou non la même qualité de service (Q(i) = Q(j) ou Q(i) Q(j)). Note that with this convention it is possible to have twice the

same block size (Xi = Xj with ij) whether or not they have the same quality of service (Q (i) = Q (j) or Q (i) Q (j)).

 In fact, it suffices for two blocks of the same quality of service to be of the same size as the channel encoder to produce a set of coded blocks having at least two elements of the same size.

SUMMARY OF X RATINGS AND Y RATING
In summary for the downlink 1,2, ..., k are indexes for all possible sizes of block to balance in flow, knowing that separately the block sizes corresponding to different service qualities. For the uplink link 1,2, ..., k are the indexes of the block list to be balanced in the rate for a given multiplexing frame.

 Y1, ..., Yk are the block sizes that correspond respectively to X1, ..., Xk after the flow balancing.

COMMON PART BETWEEN THE DOWNLINK AND THE AMOUNT
Suppose that for any quality of service q in {1, ..., p} we have the two characteristic integers Eq and Pq defined in section 0.

L'étape suivante consiste à définir le paramètre LMAX par :
LMAX = max{Lq} Ensuite on défini un entier Sq pour toute qualité de service q par :
Sq = LMAX#Eq

Sq est tel que le nombre rationnel PBASE LBASE est le rapport d'équilibrage de débit minimal étant donné le taux de poinçonnage maximal Pq/PBASE pour chaque qualité de service. The first step of the algorithm is to calculate for all q from 1 to p an integer parameter Lq defined by:

The next step is to set the LMAX parameter to:
LMAX = max {Lq} Then we define an integer Sq for any quality of service q by:
Sq = LMAX # Eq

Sq is such that the rational number PBASE LBASE is the minimum flow balancing ratio given the maximum punching rate Pq / PBASE for each quality of service.

<Desc / Clms Page number 14>

ou #x# est le plus petit entier supérieur ou égal à x. PART OF THE DESCENDING LINK SPECIFIC ALGORITHM
The relation X, -> Y, is then defined by:

where # x # is the smallest integer greater than or equal to x.

1. pour toutes les QoS q fait Lq = [ PBASE-É4)LBASE J 2. LMAX := max/q{Lq} 3. pour toutes les QoS q fait Sq := LMAX. Eq

pour i := fait Sam' X 4. 'PBASE-LBASE PARTIE DE L'ALGORITHME SPECIFIQUE AU LIEN MONTANT
Pour le lien montant les rapports d'équilibrage de débit sont calculés pour chaque trame de multiplexage. Ainsi ce n'est pas une application X1#Y1 qu'il s'agit de déterminer, mais bien plutôt une application (X1, X2, - - -, Xk) #

(Yi, Y2, ..., Yk). SUMMARY FOR DOWNLINK: @ @, (PBASE-Pg) -LBASE,

1. for all QoS q does Lq = [PBASE-É4) LBASE J 2. LMAX: = max / q {Lq} 3. for all QoS q does Sq: = LMAX. eq

for i: = done Sam 'X 4.' PBASE-LBASE PART OF THE ALTERNATIVE ALGORITHM FOR THE AMOUNT
For the up link the rate balancing reports are calculated for each multiplex frame. So it is not an application X1 # Y1 that is to determine, but rather an application (X1, X2, - - -, Xk) #

(Yi, Y2, ..., Yk).

on peut définir un ensemble {N1, ...,Nr} avec Nez... <n des charges utiles possibles pour les trames de multiplexage. In addition, the payload of a multiplexing frame can vary from frame to frame. Indeed it is possible to vary frame to frame CDMA spreading factor in the uplink link. For very high speed the payload can be further increased by a multicode transmission. Thus

we can define a set {N1, ..., Nr} with Nez ... <n possible payloads for the multiplexing frames.

Also one of the products of the flow balancing algorithm is to select a load using NJSEL and ensure that:

We proceed in two stages. In the first step, block sizes Y are determined; statically similar to the case of the downlink. It is therefore an application X, # Y; In a second step NJSEL is dynamically determined and Y, corresponding to Y;

<Desc / Clms Page number 15>

{Y, Y2,... Yk) - {Y, Y2, ..., Yk). La première étape est simplement définie par l'équation : Y; = SQ(i).X@ Ensuite on détermine JSEL par l'équation suivant :

so as to check equation (1). This is an application

{Y, Y2, ... Yk) - {Y, Y2, ..., Yk). The first step is simply defined by the equation: Y; = SQ (i) .X @ Then JSEL is determined by the following equation:

In other words, the minimum payload for transmission is selected.

Où #x# est le plus grand entier inférieur ou égal à x. Then we define the integers Zo, Z1, ..., Zk by: Z0: = 0

Where # x # is the largest integer less than or equal to x.

... c..... (PBASE-Pa)-LBASE 1. pour toutes les QoS q fait Lq := L E q 2. LMAX := max/q{Lq} 3. pour toutes les QoS q fait Sq := LMAX. Eq 4. pour i : = 1 à k fait Y; := SQ(@).Xi

Finally the Y's are simply defined by: Y, = Z, - Zi-1 SUMMARY FOR THE AMOUNT LINK

... c ..... (PBASE-Pa) -LBASE 1. for all QoS q does Lq: = LE q 2. LMAX: = max / q {Lq} 3. for all QoS q does Sq: = LMAX. Eq 4. for i: = 1 to k makes Y; : = SQ (@). Xi

7. pour i : = 1 à k fait Y, := Z, - Z@-1

7. for i: = 1 to k makes Y,: = Z, - Z @ -1

<Desc / Clms Page number 16>

Lastly, although the notion of quality of service was defined as the quality of service of a transport channel, ie the quality of the service offered by Level 1 at higher levels, it would be more accurate, with regard to the determination of the rate-of-flow balancing, to speak of the quality of service offered by the bottom of the interleaving and multiplexing chain to the channel coder.

données d'entrée : Xi -- nombre de bits en entrée Yi -- nombre de bits en sortie Npr=Y;-X; -- nombre de bit à répéter ou à poinçonner (si Y,>Xi on répète, sinon on poinçonne La règle de poinçonnage/répétition est comme suit : e = 2*Np/r - Xi -- erreur initiale entre le rapport courant et le rapport désiré de poinçonnage/répétition x = 0 -- index du bit courant tant que x< X, fait si e > 0 alors -- teste si le bit numéro x doit être répété/poinçonné poinçonne ou répète le bit numéro x "S e = e + (2*Np/r - 2* Xi) -- mise à jour de l'erreur Sinon Sinon e = e + 2*Np/r -- mise à jour de l'erreur fin~si x = x + 1 -- bit suivant fin fait Tableau 1 Algorithme de répétition ou de poinçonnageThe notion of quality of service used in the flow balancing algorithm therefore corresponds to: # a reference point between the channel coder 104 and the first interleaver 130 in the upstream link chain of the figure 1, and e # at a reference point between the channel encoder 104 and the rate balance 124 in the downlink

input data: Xi - number of input bits Yi - number of output bits Npr = Y; -X; - number of bits to be repeated or punched (if Y,> Xi is repeated, otherwise it is punched The punching / repeating rule is as follows: e = 2 * Np / r - Xi - initial error between the current ratio and the desired ratio of punching / repetition x = 0 - index of the current bit as long as x <X, does if e> 0 then - test if the bit number x must be repeated / punched punch or repeat bit number x "S e = e + (2 * Np / r - 2 * Xi) - update of the error Otherwise Otherwise e = e + 2 * Np / r - update of the end error ~ if x = x + 1 - bit following end done Table 1 Algorithm of repetition or punching

Claims (6)

1.- telecommunication system in which several qualities of service can be multiplexed on the same link, and are advantageously balanced in Eb / l characterized in that the entities at both ends of the link are adapted to exchange information which can be deduced two parameters (Eq and Pq) for each quality of service (q), the proportion between the first parameters (Eq) fixing the proportion of Eb / l between the different qualities of service, while the second parameters define a maximum punching rate for the corresponding quality of service.  2. - System according to claim 0 characterized in that two entities of the system establishing a communication link, or modifying a communication link which involves a multiplexing of several qualities of service are adapted to proceed according to a protocol comprising the steps in which: the first entity signals to the second its maximum transmission capacity, or this capacity is implicitly known by the second entity by other means, the second entity determines a proportion of Eb / l that it expects in reception, and which is possible given the maximum transmission capacity of the first entity, and derives therefrom a corresponding list of values for said first parameters (Eq) - the second entity indicates the proportion of Eb / l that it expects in reception at means of said list of values.  3. - System according to any one of the preceding claims characterized in that said parameters are bounded integers, such that the first is positive non-zero, and the second positive or zero.  4. - System according to any one of the preceding claims characterized in that it is a mobile radio system CDMA type.  5. - System according to any one of the preceding claims, characterized in that if Eq and Pq designate respectively said first and second parameters for a given quality of service q, and if X1, X2, ..., Xk is the list of flow-balanced block sizes for a given quality of service when this quality of service describes all the values pos- <Desc / Clms Page number 18>  for all qualities of service q does 1 (PBASE-Pg) -LBASE - for all qualities of service q does Lu: ~ E q F- - LMAX: = rrx {Lq} for all qualities of service q does Sq: = LMAX. Eq for i: = does SOM 'XI for i: = 1 at k makes Y,: = rPBASELBASE

 sibles for the link in question, then the list of sizes Y1, Y2, Yk corresponding blocks balanced flow is determined by the sequence of the following steps: 6. - System according to any one of the preceding claims, characterized in that if Eq and Pq respectively designate said first and second parameters for a given quality of service q, if N1, N2, ..., Nr is the list of charges. useful multiplexing frames, if the blocks to balance in rate for a given multiplex frame are indexed from 1 to k, and if X1, X2, ..., Xk is the list of their respective sizes, then the list of sizes Y1, Y2, Yk corresponding flow-balanced blocks is determined by the sequence of the following steps:

 for all the qualities of service q does L (PBASE-Po) .LBASEJ - for all qualities of service q does Lq L p for service Eq - LMAX: = max / q {Lq} - for all qualities of service q do Sq: = LMAX. Eq - for i: = 1 at k makes Y; : = SQ (@). X @

 i = k JSEL: = min j /, 2: V; ::; PBASE.LBASE.N 1 = 1 Y;} NJSEL for i: = 1 at k makes bzz 1 Y 'J = l - for i: = 1 at k makes Y,: = Zi- Z @ -1.