WO2006070135A1

WO2006070135A1 - Method and equipment for allocating resources of a telecommunication cellular network for mobile terminals

Info

Publication number: WO2006070135A1
Application number: PCT/FR2005/003274
Authority: WO
Inventors: Jérôme GALTIER
Original assignee: France Telecom
Priority date: 2004-12-24
Filing date: 2005-12-23
Publication date: 2006-07-06
Also published as: FR2880231A1

Abstract

The invention concerns a method for allocating resources to a set of mobile telecommunication terminals of a cellular network so as to maximize an overall utility function corresponding to the sum of the utility functions particular to each terminal while satisfying constraints related to network resources provided to the terminals, said method consisting in: assigning minimum and maximum resource values for each terminal; computing for each terminal a function representing resources consumed by the terminal; and assigning an optimal value of consumed resources for each terminal based on the values of said computed function.

Description

METHOD AND EQUIPMENT FOR ALLOCATING RESOURCES

A CELLULAR TELECOMMUNICATION NETWORK

FOR MOBILE TERMINALS

The invention generally relates to mobile-type telecommunication networks, and more particularly relates to the allocation of the available resources of a cellular telephone network to mobile terminals of a network telephone. By allocation of resources means the provision of the means proposed by the network to allow a terminal located in a cell of the network to exchange data.

Thus, a particularly advantageous application of the invention relates to the optimal and fair allocation of rates to mobiles of a telecommunication network.

In fact, telecommunication network operators generally seek to maximize the number of users that can be served by a given network. In CDMA-type telephony networks, for a given user, the other users, in the upstream direction, or the base stations other than those with which it communicates, in the downstream direction, constitute sources of noise which disturb the user. If the number of users becomes too large, the noise increases and causes a drop in quality, or even a blocking of communication. Mobile data rate control can control and limit this noise and thus improve performance.

It is for these reasons that one of the constant concerns of telecommunication operators is to optimize the management of mobile speeds and optimize the admission of new terminals in the cells so as to maximize the number of terminals in each cell, while ensuring a given quality of service (QoS). Generally, the algorithms implemented to control the data rate of the terminals in the cells as well as to control the admission of new terminals are based on the control of the charge level. For example, if the admission of a mobile into a cell causes an increase in the charge level such that the total charge exceeds a given threshold, the incoming mobile is forced to reduce its rate to be admitted. Beyond a certain threshold, all new incoming terminals are rejected.

Thus, in the state of the art, when the load of a cell is too large to allow the admission of a new terminal, the authorized bit rate for this terminal is reduced, which causes a drop in the quality of service for the incoming terminal. Terminals already in communication are not affected.

Admission policies for new terminals can be implemented in a variety of ways. For example, it is possible to adapt the flow of the incoming mobile by guaranteeing a given rate. This implementation induces relatively large rejection rates and therefore limits the capacity of the network. It is also possible to adapt the rate of the incoming terminal so as to always admit it, even with a zero flow. This implementation does not guarantee the quality of service (QoS). Finally, these various strategies are not optimal, to the extent that it is possible to increase the speed of a mobile without necessarily reducing that of another.

Thus, conventional flow allocation techniques do not allow to allocate rates to terminals in a network optimally and equitably. They generally do not allow exploring the full range of equitable strategies for dynamic management of mobile bit rates in the network. Nor do they allow the full range of mobile admission management strategies to achieve a rejection rate between extreme values, one for guaranteed quality of service and one for guaranteed quality of service. unsecured quality of service. To overcome these drawbacks, it has been proposed to calculate the bit rates allocated to each terminal, when a new terminal enters a cell, so as to maximize a utility function. Various techniques can be used to solve such an equation. Calculation algorithms using Lagrangian relaxation, positive semi-definite programming or linear / whole mixed programming can be used to solve such an equation. Maximizing a utility function to allocate resources to terminals in a network involves the use of additional constraints related to network resources. For example, the resources assigned to users can not be less than the respective minimum threshold values assigned to users. Similarly, they can not be greater than respective maximum threshold values assigned to the users. Finally, the sum of the resources allocated to the users can not be greater than the overall capacity of the system. However, known calculation techniques used to maximize the utility function taking into account these constraints all have the disadvantage of being very slow in computing time and limited reliability. However, the use of equations to maximize the utility function while taking into account constraints related to network resources is based on real-time constraints and reliable responses. In practice, it is a matter of determining what will be a new bit rate allocation each time a mobile enters the network, in real time.

In view of the foregoing, the object of the invention is to make it possible to allocate resources of a telecommunication cellular network to mobile telecommunication terminals located in the network in real time and reliably.

The subject of the invention is therefore a method for allocating resources of a telecommunication cellular network for a set of mobile telecommunication terminals of the network, according to which the resource is allocated to the terminals so as to maximize a global utility function. , estimated at from functions of utilities specific to each terminal, while satisfying constraints related to the network resources made available to the terminals.

According to a general characteristic of the invention, the resource allocation method comprises the following steps: assignment of a utility function (fi) and minimum (mi) and maximum (Mj) resource values to each terminal ;

- determining a function (fj ¹ ) representative of the resources consumed by each terminal; and

determining an optimum value (λ _t ) of resources for each terminal from values of said functions representative of the resources consumed, and

- Assigning said optimum value to the terminal for which this value has been determined.

In one implementation mode, the global utility function is the sum of the utility functions specific to each terminal.

In an alternative, the global utility function is the product of the utility functions specific to each terminal. However, when implementing the method according to the invention, it is possible to reduce to the previous case, by using the properties of the logarithm function and application of the method described in more detail below to the logarithm of the functions of utility. ". In one embodiment, to calculate the representative function of the resources consumed by each terminal,

the value of the derivatives of the utility function proper to each terminal is calculated for each minimum and maximum value; the calculated values of the derivative of the utility function specific to each terminal are sorted;

an interval is searched in which is located a pivot value of the calculated values which corresponds to an optimal value of resource; and,

in the interval sought, the pivot value is calculated, the optimum value assigned to a terminal corresponding to the value of minimum resource, either at the maximum resource value or at a value calculated from the pivot value.

For example, the calculated resource value {λ _; ) from the pivot value is calculated, for terminals i for which the pivot value is used to determine the optimal value, from the following relation:

where (Z ¹ ,) ^"1 represents the inverse of the derivative of the utility function and /? * denotes a real dual coefficient.

Similarly, for example, the calculated values are sorted by increasing values.

According to another characteristic of the invention, the constraints related to the resources of the network require that a resource value assigned to a terminal must be greater than the minimum resource value.

In addition, network resource constraints dictate that a resource value assigned to a terminal must be less than the maximum resource value.

Finally, the sum of the resource values assigned to the terminals must be less than the overall capacity of the system.

According to yet another characteristic of the invention, the representative function of the resources consumed by each terminal is determined from the utility functions of all the terminals. For example, the function representative of the resource consumed by each terminal i is given by the relation:

φ (y) = Σm _i + ΣM _i + Σ (ff ⁱ⁾ (y)

in which :

Σm, corresponds to the sum of the resources allocated to the terminals to which the minimum resource value is affected;

ΣM, is the sum of the resources allocated to the terminals to which the maximum resource value is assigned;

( ^υ Y ¹ > Y '^"1 represents the inverse of the derivative of the utility function / ..

In addition, the pivot value that corresponds to the optimal values of the resources can be given by the relation:

_/? * = [Σ (/'.)- ¹ ] ^"1 (C-Σm ,. -ΣM ,.)

where C denotes the overall capacity of the network.

The subject of the invention is also a method for managing the resources of a telecommunication cellular network for a set of mobile telecommunication terminals of the network, in which, at each new terminal entering the network, an overload detection is carried out. of a cell of the network and, in the case where one of the cells of the network is overloaded, a resource allocation is carried out by implementing the allocation method as defined above, to accept the incoming terminal.

According to a third aspect, the invention relates to a resource allocation equipment of a cellular telecommunication network for a set of mobile telecommunication terminals of the network, the equipment suitable for implementing the method according to the invention. and including,

means for assigning a utility function (ή) and minimum (mi) and maximum resource values (MO at each terminal (T);

means for estimating a ^utility ^function 9 ^to ^perform utility ^functions specific to each terminal,

means for determining a function (f ¹ ) representative of the resources consumed by each terminal; means for determining an optimum value

resources for each terminal from values of said functions representative of the resources consumed, and

means for assigning said optimum value to the terminal for which this value has been determined.

This equipment notably comprises data processing means able to implement the different steps of the method according to the invention. Finally, according to a fourth aspect, the invention relates to a software module recorded on a medium, characterized in that it comprises instruction codes for the execution of a resource allocation method as defined above. above.

Such a data medium may be a hardware storage medium, for example a CD-ROM, a magnetic diskette or a hard disk, or a mobile medium such as an electrical, optical or radio signal.

Other objects, features and advantages of the invention will become apparent on reading the following description, given solely by way of nonlimiting example, and with reference to the appended drawings in which: FIG. 1 schematically illustrates the architecture a telecommunication cellular network implementing a method according to the invention;

FIG. 2 is a diagram illustrating the main phases of a resource management method of a cellular network according to the invention;

FIG. 3 is a table showing examples of resource values assigned to terminals;

FIG. 4 illustrates the main phases of the flow allocation method according to the invention;

FIG. 5 is a table illustrating the calculation of the assigned resource values; and FIGS. 6 and 7 show curves illustrating the evolution of the resources consumed.

In Figure 1, there is shown the general architecture of a cellular mobile network implementing a resource allocation method according to the invention.

This method is intended to be applied to the network of CDMA or W-CDMA type.

In this figure, only three cells have been represented, for the sake of clarity. As we can see, the territory to be served is broken down into cells

Q, C ₂ , and C ₃ . A cell is linked to a base station BS (or "node B"), which has an antenna for transmitting to terminals, such as T, and receiving signals from them.

As seen in the figure, the base stations are managed by a radio resource controller or RNC ("Radio Network Controller" in English) that manages the radio interface. In particular, the RNC controller comprises all the hardware and software means and is properly programmed to allocate resources to each terminal of the Q, C ₂ and C ₃ cells of the radio subsystem which it manages. For this purpose, it comprises in particular the hardware and software means and is programmed to calculate a function which represents the resources consumed by the terminals of the network and to allocate optimal values of resources, in particular of the bit rate, from this function, as will be discussed later.

This allocation is an optimal allocation of resources, which results in the renegotiation of the resources allocated to each terminal in order to obtain a maximum resource value for the terminals as it can be used. It is no longer possible to increase the resources allocated to one of the terminals without having to reduce the resources of one or more other terminals. For example, it is a question of renegotiating the flows in order to optimize the capacity of the network.

According to one characteristic of the invention, the calculation of the resources goes through the maximization of a utility function. In particular, it is to maximize the following function:

It is furthermore necessary to satisfy the following constraints: λ -m, ≥ 0 V,, P) (2)

M-X. ≥ 0 V, - e {l, .. -F} (3)

In these relations (1) to (4):

X ₁ is a variable in the set of positive reals and denotes resources allocated to the terminal i,

- m ,, is a datum in the set of positive reals and denotes a minimum resource bound assigned to the terminal i, - M ,, is a datum in the set of positive reals and denotes a maximum resource bound assigned to the termina ! i

- It is a datum in the set of positive reals and denotes the overall capacity of the system, and

- /, is a function of the set of positive reals in the set of positive reals and denotes a utility function associated with the resources for the terminal i.

Moreover, in these relations (1) to (4), we assume that p is a positive integer, that the function /, is a concave function, and that C, m,, M ₁ are positive reals for i belonging to (1, ..., p). The constraints defined by relations (2) to (4) require that a resource assigned to a terminal must be between a minimum terminal m, and a maximum terminal M ₁ , and the sum of the resources allocated to the terminals must remain less than the overall capacity C of the system.

Thus, with reference to FIG. 2, at each new input of a terminal in a cell of the network (step 10), it is detected whether there exists in the network an overloaded cell (step 12) in a manner known per se. If this is not the case, the terminal is accepted (step 14). On the other hand, if there is an overloaded cell, allocation of the resources of the network is carried out by reallocating the rates of mobiles to the cell (step 16). If, during the next step 18, the calculation of the reallocation of the flows makes it possible to complete, the procedure returns to the previous step 12. Otherwise, the mobile is refused

(step 20).

We will now describe the procedure for allocation of resources to the terminal, when a cell is overloaded. As indicated above, the overloading of a cell can be detected during the admission of a mobile, that is to say, in general, when a terminal enters a cell, for example when the establishment of a call or when passing a terminal from one cell to another.

This resource allocation procedure is implemented within the base station controller which comprises all the appropriate hardware and software means to implement this procedure.

Such a procedure involves calculating a new rate value for the terminals of the cell to maximize the following global utility F function:

which corresponds to the sum of the utility functions f. {λ) assigned to each terminal i. These functions of individual utilities // characterize economic or private interests to receive, for a termina! i, a resource λ .. Thus, the global utility function makes it possible to take into account all the terminals. However, this procedure requires above all compliance with the aforementioned constraints related to the operation of the cell. In particular, it is therefore necessary to check that the load of each station, in particular that of the station in which the mobile terminal enters, does not exceed a maximum permissible load. Moreover, the power demanded at a set of base stations, in particular at the station of the cell in which the mobile enters, must not exceed a maximum permissible power. Finally, the sum of the resources allocated to the terminals must not exceed the overall capacity of the system.

Thus, the constraints to be satisfied correspond to the relations (2) to (4) mentioned above.

Note that in a cell, each terminal receives a resource A ₁ . bounded according to the values of the table of FIG. 3. In other words, each resource value A ₁ lies between the threshold values m ₁ and M ₁ . In the realization set considered, a utility coefficient β _t . is allocated to each terminal for calculating the utility function fj (x). For example, in the case illustrated in the table of FIG. 3, the total capacity C of the system is equal to 100. Before the arrival of the terminal 4, the cell is not saturated and each mobile terminal i receives its maximum demand flow rate M,.

We will now describe, with reference to Figure 4, the resource allocation procedure implemented in step 16 previously mentioned. This procedure consists first in affixing to each terminator a utility function ft and the values of the minimum and maximum resources m,

and M ₁ (step 22). In particular, the utility function fi can be expressed as form:

in which "denotes a strictly positive real coefficient different from 1.

The value of the derivative of the utility function is then calculated for the values m ₁ and M ₁ for each terminal i (step 24).

In the next step 26, the values calculated in the preceding step 24 are sorted by increasing values.

For this purpose, a conventional fast sorting algorithm is used to obtain the sorted list of values. We obtain at the output a list which gives for each k belonging to the interval {l, ..., 2p] the following information on the k-th value:

User (k) = i Terminal (k) = m _t . or M ₁ .

Value (k) = fi (Terminal (k)). We will have Value (1) ≤ Value (2) ≤ ... ≤ Value (2p) and for each terminal i, there are two values ki <k ₂ with User (k-ι) = User (k ₂ ) = i, Terminal (ki) = M. and Terminal (k ₂ ) = m,. Note in this case Start (i) = (ki) and End

(i) = (k ₂ ).

In the next step 28, the interval of the pivot value is determined, which corresponds to an optimum value of the flow rates.

This value corresponds practically to a value which is used for the optimization of the resources and which gives a global index of satisfaction of the terminals. It is calculated according to the capacity that can be provided to the terminals. To locate the interval of the pivot value we use the previous list and we search by dichotomy. We then assign the values Low = 1 and High: = 2p, and we apply the following algorithm: As High - Low> 1 Medium: = (High + Low) / 2 Phi: = O

For all terminals i If Start (i)> Middle phi: = phi + M ₁ . If Fin (i) <Middle phi: = phi + m.

If Start (i) <Middle and End (i)> Middle phi: = phi + (f ι) ^"1 (Middle)

If phi> C then Low: = Middle if High: = Middle Out, we provide the value of High and Low, with High-Low = 1 and {High, Low} included in the range {l, ..., 2p). In the following step, the actual value of the pivot value p is calculated from the relation:

p * = ψ- ^ι (Keste) Where ψ ≈ Σ / r (6) i

The value Rest is calculated in particular from the following algorithm

Rest: = C For all terminals i

If Start (i)> Low Remain: = Remain - M ₁ If End (i) <High Remain: = Remain -m. ψ then designates the sum of the functions (//) ^" 'for i checking Start (i)

<Low and High (i)> High.

Finally, the calculated values are assigned (step 31). For each terminal i, the resource value λ. assigned to it is given by:

If Start (i)> Low, then λ,: = M,

If End (i) <High, then À,: = m. Otherwise, V = (/ ', TV) -

As will now be demonstrated later, the resources allocated to the terminals thus calculated constitute an optimal resource allocation solution. Suppose the functions fi, for i included in the interval

{l, ..., p] are strictly concave, that the constraint:

is checked at equality by an optimal solution, and that the function fj is computable in one operation. Suppose that we can calculate the inverse of any sum of inverses of f'i for i included in the interval {l, ..., p] in M operations. Then the resources, that is to say His solution to the relations (1) to (4), are computable in M + O (p log (p)) operations, where O denotes the Landau notation.

By applying the Kuhn-Tucker conditions, we obtain, for the optimal solution A ^* for i belonging to the interval {l, ..., p), and the real positive coefficients duaux pi, v _t for i included in { l, ..., p} and p, rM) -μ, + v, + p ≈ 0 V, e {l,. ", />},

//, (4 -ιn,) = 0 V ,. e {l, ...,;>},

V ₁ (M ₁ - ^) = OV ₁ e {1, ...,>}, βt

For each i in the interval {l, ..., p), three cases may occur.

1) λ ^* -m. In this case v, = 0. Therefore - / ',. (A ^* ) - // ,. + p = 0, which gives / '. (λ ^* ) ≤ p.

2) λ ^* -M ₁ . Incecas μ, = 0. Let's say that - / ',.(λ ^* ) -v, + p = 0, which is /'.(λ ^* ) ≥p.

3) λ ^* e] m., M ₍ . [. We then have v. = 0 and μ _{ . - 0, ie rM) ^{≈ p} - In other words, if /. ^Is strictly decreasing and continuous with i belonging to the interval {l, ..., p], it is invertible and each actual value of p, one can associate a capacity "consumed" as defined by:

As we understand it, φ is a decreasing function of p. An optimal solution is then given by a value p * which verifies:

<p (p *) = C (8)

The algorithm for calculating the solution in relation (I) to (4) above can be implemented as follows.

Let X = {mm _p } U {Mi, ... M _p }. We have \ x \ ≤ 2 _p . We first calculate values of the functions f. to obtain the set Y defined by the relation:

Y = {/ '. {M). , ie [l, p] (u {/ ',. (M ₁ ) J _€ [l, p]} The set Y is then sorted. This sorting is done in O (p log (p)) operations. We then reason by dichotomy on Y. For a value y <= Y, one computes φ (y) in O (p) operations, and in O (log (p)) stages one finds two consecutive values yi <y ₂ , such that φ (y _x ) ≥ C and φ (γ ₂ ) <C. In borderline cases, V2 is 0, where yi is infinite.

If ç (y-ι) = C, then the equation p * = yi gives a solution.

Otherwise, the network resources consumed by a terminal i are given by:

vve] y ,, _Λ l M _{1 +}

(FiViy).

, ': /; (Λ /,) ≤ _Λ < _Λ ≤ /; (/ n,)

The pivot value p * is then given by the following relation:

p * = kf _ι '(M _i ) ≤y _ι <y ₁ ≤f _l (m _l ) ur ^{c ~} Σ i - Σ ^M <O)

'-yi> fl {m,) iy [<fi ( ^M ι)

The utility function fj (x) assigned to the terminals can be of various forms. l-α

In the case where f _i (x) = β. , for α e] θ; l [l)] l, + ∞ [ _> ^then l - α the allocation of resources is computable in O (plog (p)).

In this case, f! (X) = β _i x ^{~ a} and fi ^{~ ι} (y) ≈ - y ^{~ Ua} - So, for ri

J e {l, ..., p} if we ask

so

For J given, this function is computable in O (p) operations. In the case where fj (x) = ln (x), then the allocation of resources is computable in O (p log (p)).

In this case, f ^' ( _x ) = - and f- ^{~ {} {y) = - - So, for J e {l, ..., p] if we xy pose

i <= J

so

Finally Yes

, then the solution is computable in O (p log (p)).

In this case, f _t ^' (x) = β _r By reasoning in a similar way to the result above, we consider Y = {β,: ie {1, ..., p}}, and

φ is a descending stair function. We look for two consecutive values yi and y ₂ of Y, such that yi <y ₂ , with #> (yi) ≥C and ç? (Y ₂ ) <C. In this case, an optimal resource Z ^* is given by:

Concretely, and referring now to FIGS. 5-7,

Λ-a will first consider the case where f _s (x) = P ₁ for "= 1/2,

\ -a f- (χ) = β, l4χ. The calculation of the optimal resource values is done, as indicated previously, in several steps.

It is first necessary to calculate the value of the derivative of the utility function for each minimum value and each maximum value of resources, and for each terminal i, the values visible in the table of FIG. 5 are then obtained. The values of this array are then sorted in O (p log (p)), and we get:

U (M ₄ ) = K (M ₃ ) <W (M ₁ ) <f ₄ ^' (m ₄ ) = f ₃ ' (m ₃ ) <Mm ₁ ) <V (M ₂ ) <f ₂ '(m ₂ )

The function φ is described as follows: for x ε [0.91; 1, 58] φ (x) = 60 + 50 / x ² ; for x ε [1.58; 2.24]; (x) = 20 + 150 / x ² ; for x € [2.24; 3.16] p (x) = 30 + 100 / x ² ; for x € [3.16; 4.47]; (x) = 40; and for x € [4.47; 8.94] p (x) = 20 + 400 / x ² .

This function is shown in FIG. 6. The aim is to identify the position where φ (x) = 100. The pivot value is then in the range [0.91; 1, 58], where the function φ is given by φ (x) = 60 + 50 / x ² , which gives a pivot for x = 0.73. The bit rates obtained will therefore reduce the terminals 3 and 4 to the common value of 20.

Likewise, referring again to FIG. 5, in the case where fj (x) = 1n (x), the calculated values for f are as follows:

W (M ₁ ) <f ₃ (M ₃ ) = W (M ₄ ) <f ₂ '(M2) <Mm ₁ ) <W (In ₂ ) = f ₃ ' (m ₃ ) = f ₄ '(m ₄ ) .

The function φ plotted in FIG. 7 is described as follows: for x e [0.025; 0.033] φ (x) = 80 + 1 / x; for x e [0.033; 0.050] φ (x) = 20 + 3 / x; for x e [0.050; 0.100] φ (x) = AIx; and for x e [0.100; 0.200] ^ (x) = 10 + 3 / x, which gives a range for the pivot value of [0.033; 0.050], and a final value of 0.0375.

Finally, if fj (x) = ιτijX, the values of f can be sorted as follows:

f ₃ '(m ₃ ) = K (M ₃ ) = U' (m ₄ ) = W (M ₄ ) <W (Ui ₁ ) = W (M ₁ ) <f ₂ '(m ₂ ) = W (M ₂ ).

The function φ (x) is then a function constant at intervals given by: for x e [5; 10] "φ (x) = 70; and for x e [10; 20] φ (x) = 40, the pivot value is then in the range of [5; 10].

Claims

1. Method for allocating resources of a cellular network (Ci, C2,

C ₃ ) for a set of mobile telecommunication terminals (T) of the network, the resources being allocated to the terminals so as to maximize a global utility function estimated from utility functions specific to each terminal, while satisfying constraints related to the network resources made available to the terminals, characterized in that it comprises the following steps:

assigning a utility function (fj) and minimum (mi) and maximum (Mi) resource values to each terminal (T);

determining a function (fi ') representative of the resources consumed by each terminal;

determining an optimum value vΛ) of resources for each terminal from values of said functions representative of the resources consumed, and

- assignment of said optimum value to the termina! for which this value has been determined.

The method of claim 1, wherein the global utility function is the sum of the utility functions specific to each terminal.

3. Method according to one of claims 1 to 2, characterized in that it comprises the steps of:

calculating the values of the derivative of the utility function specific to each terminal for each minimum and maximum value;

sorting the calculated values of the derivative of the utility function specific to each terminal; searching for the interval in which a pivot value of the calculated values is located which corresponds to an optimal value of resource; and

calculating, in the sought interval, the pivot value, the optimal value assigned to a terminal corresponding to either the minimum resource value or the maximum resource value, or to a value calculated from the pivot value.

4. Method according to claim 3, characterized in that the calculated resource value (A ₁ ) from the pivot value is calculated, for the terminals i for which the pivot value is used to determine the optimal value, from the next relationship

in which (/ j 'represents the inverse of the derivative of the utility function and p * denotes a real dual coefficient.

5. Method according to one of claims 3 and 4, characterized in that the calculated values are sorted by increasing values.

6. Method according to any one of claims 1 to 5, characterized in that the constraints related to network resources require that a resource value assigned to a terminal must be greater than the minimum resource value.

7. Method according to any one of claims 1 to 6, characterized in that the constraints related to the resources of the network require that a resource value assigned to a terminal must be less than the maximum resource value.

8. Method according to any one of claims 1 to 7, characterized in that the sum of the resource values assigned to the terminals is less than the overall capacity of the system.

9. Process according to any one of claims 1 to 8, characterized in that the function representative of the resources consumed by each terminal is determined from the utility functions of all the terminals.

10. Method according to claim 9, characterized in that the function φ (y) representative of the resource consumed by each terminal i is given by the relation:

_< p (y) = ΣOT _/ + ΣΛ / _/ + Σ (/ - O ^{(~ l)} (y) where: Σm, is the sum of the resources allocated to the terminals to which the minimum resource value is assigned ;

IM ₁ is the sum of the resources allocated to the terminals at which the maximum resource value is allocated; and

(fi) ¹ represents the inverse of the derivative of the utility function fj.

11. Method according to claim 10, characterized in that the pivot value corresponding to the optimal values of the resources is given by the relation:

where C is the overall capacity of the network.

12. A method for managing the resources of a telecommunication cellular network (Ci, C ₂ , C ₃ ) for a set of mobile telecommunication terminals (T) of the network, characterized in that at each new terminal entering the network. , an overload detection of a cell of the network is carried out and, in the case where one of the cells of the network is overloaded, a resource allocation is carried out by implementing the allocation method according to one of the any of claims 1 to 10 to accept the incoming terminal.

13. Equipment for allocating resources of a cellular network of telecommunication system for a set of mobile telecommunication terminals of the network, characterized in that it comprises

means for assigning a utility function (fi) and minimum (nrij) and maximum (Mj) resource values to each terminal (T); means for estimating a ^utility ^function 9 ^to ^perform utility ^functions specific to each terminal,

means for determining a function (f ¹ ) representative of the resources consumed by each terminal;

means for determining an optimum value vΛ / of resources for each terminal from values of said representative functions of the resources consumed, and

14. Software module recorded on a medium, characterized in that it comprises instruction codes for the execution of a resource allocation method according to any one of claims 1 to 12.