A method for admission control for mobile networks, an admission controller and a communication system therewith
The invention is based on a priority application EP 05290415.8 which is hereby incorporated by reference
Field of the invention
The invention relates to a method for providing admission control for the admission of a mobile terminal to a mobile network according to the preamble of claim 1, an admission controller according to the preamble of claim 6 and a communication system according to the preamble of claim 8.
Background of the invention
The IEEE 802.lie standard provides Quality of Service (QoS) capabilities to Wireless Local Area Network (WLAN) systems by supporting traffic differentiation by means of Enhanced Distributed Channel Access (EDCA) mechanism, time-division multiplexing by means of Hybrid coordination function Controlled Channel Access (HCCA) mechanism, and means to request, release or modify network resources. The set of IEEE 802.lie tools promises to allow or ease the deployment of QoS-sensitive realtime and interactive services like Voice over Internet Protocol (VoIP) telephony and video streaming.
On the other hand, also admission control is a key building piece for QoS support. It is used to control network resources for QoS-constrained traffic. The admission control function accepts or rejects resource reservation requests based on network resource availability.
Unfortunately, while defining the messages to request, release or modify network resources, the IEEE 802.lie standard does not define a complete admission control mechanism, comprehensive of decision algorithm and means to estimate channel utilization.
Currently there are only either best effort solutions, which provide no QoS support, like in legacy IEEE 802.11a/b/g WLAN systems, or non-controlled solutions based on traffic differentiation using e.g. EDCA mechanisms, for the access to the network resources available.
In case of solutions based on traffic differentiation like e.g. EDCA mechanisms, the network resources are used by the terminals without control. The use of priorities brings effective traffic differentiation (e.g. voice with higher priority than web traffic), but the lack of resource control does not protect such solutions from network, i.e. radio channel, saturation. The main problem is, that a newly activated traffic flow, e.g. a voice call, can impact all already existing (active) streams like e.g. voice calls by degrading the overall satisfaction of the full set of users/customers of the WLAN system.
The object of the invention is to propose a solution for providing admission control for the admission of a mobile terminal to a mobile network.
This object is achieved by a method according to the teaching of claim 1, an admission controller according to the teaching of claim 6 and a communication system according to the teaching of claim 8.
The main idea of the invention is to use a measurement-based admission control that can be applied e.g. to IEEE 802.lie EDCA-enhanced WLAN systems. The proposed solution can be combined with the QoS capabilities introduced by the IEEE 802.lie standard, that provide to mobile terminals the means to reserve and release radio resources as well as to modify them. The solution is based on the measurement of some important metrics characterizing the radio channel, as specified e.g. by the IEEE 802.11k/Dl.2 draft standard.
The resource reservation request, i.e. the so-called ADDTS request message (ADDTS = Add Traffic Stream) sent from a mobile terminal to an access point, contains traffic specifications like e.g. the nominal MSDU size (MSDU = MAC Service Data Unit), the mean data rate or the minimum PHY rate (PHY = Physical Layer).
The admission control decision is concentrated into a functional admission controller, responsible to execute the Channel Admission Control (CAC) algorithm. The admission controller can reside e.g. in a so-called fat access point or in a Wireless Network Controller (WNC). In the following the admission controller will be described as external to the access point, in order to explicitly detail the communication between the access point and the admission controller. In case of colocated access points and admission controller, such communication is internal .
The CAC algorithm takes as input the traffic specifications (TSPEC) describing the to-be-admitted flow, i.e. the so-called TSPEC element, combines it with the measured network load information, i.e. the current channel utilization, and decides whether or not the flow can be admitted without impacting the performance of already admitted flows.
Briefly, the channel admission control process can be described in the following way:
The access point measures the channel utilization U during
meas —'
the time period of a so-called measurement window (MW). Then, it provides this information to the admission controller, as input for the CAC algorithm.
The admission controller uses the channel utilization and has initial values to build its own internal view of the channel utilization that is called current channel utilization £7curr:
U => U
meas curr
When the admission controller receives a resource reservation request, i.e. an ADDTS request message, from a mobile terminal, it computes the increase in utilization of the channel requested by the new flow [7add as the synthesis of the flow characteristics specified in the TSPEC element:
TSPEC => U ,.
add
The admission controller combines the current channel utilization l/curr with the calculated increment of the channel utilization that is called additional channel utilization L7add to get the (theoretical) new channel utilization Unev, as if the new flow was already admitted:
new curr add
The admission controller takes the admission decision by checking if Unetf does not exceed a given target channel utilization U et, i.e. the maximum acceptable channel utilization. If Unew is smaller than Utarget, the new flow is admitted, otherwise it is rejected:
Vm < Utarget ? => Decision
When a flow is admitted, the admission controller updates its internal view of the channel utilization, i.e. the current channel utilization Ucurr, in order to account for the increase in the channel utilization due to the newly admitted flow:
U = U
curr new
urther developments of the invention can be gathered from the dependent claims and the following description.
In the following the invention will be explained further making reference to the attached drawings.
Fig. 1 schematically shows a communication system with several mobile terminals, two access points and one admission controller for carrying out a method for providing admission control for the admission of a mobile terminal to a mobile network according to the invention.
Fig. 2 schematically shows the principle mechanism of the CAC algorithm that is used to carry out a method for providing admission control according to the invention.
A communication system according to the invention is depicted in fig. 1 and comprises a mobile network that in turn comprises at least one access point API, at least two mobile terminals STAl and STA2 and at least one admission controller AC. In the example described hereafter by means of fig. 1, there are comprised two access points API and AP2 and four mobile terminals STAl - STA4. Preferably, said communication system additionally comprises at least one connection to a network NW like e.g. the Internet or another mobile network.
The access points API and AP2 are connected to each other e.g. via a backbone system and at least the mobile terminal STAl is connected to the access point API via a wireless connection. Additionally, the access points API and AP2 are both connected to the access controller AC via a fixed or wireless connection. The mobile terminal STAl that is connected via a wireless connection to the access point API can by means of the backbone system be further connected via the further access point AP2 to one of the mobile terminals STA3 or STA4 within the same mobile network. Furthermore, this mobile terminal STAl can also be connected by means of the backbone system and via gateways to devices like e.g. terminals or servers located in a further network NW like e.g. the Internet or another mobile
network.
The access points API and AP2 comprise the functionality of an access point of a mobile network, i.e. they provide the possibility for mobile terminals to get connected to a mobile network. Furthermore, the access points API and AP2 comprise means for sending admission requests that are originated by a mobile terminal STAl - STA4 to the admission controller AC and means for receiving admission responses from the admission controller AC.
The mobile terminals STAl - STA4 comprise the functionality of a mobile terminal for a mobile network, i.e. they can be connected to a mobile network by means of an access point API or AP2. Particularly, the mobile terminals STAl - STA4 comprise means for sending admission requests to the access points API or AP2 and for receiving admission responses from the access points API or AP2.
The admission controller AC according to the invention comprises means for receiving admission requests for the admission of a mobile terminal STAl - STA4 to the mobile network, for receiving parameters that characterize the radio channel that shall be used for a connection of the mobile terminal STAl -STA4 to the mobile network, for receiving parameters that characterize the traffic flow of said connection, for controlling the admission based on at least one of said parameters that characterize the radio channel that shall be used for a connection of the mobile terminal STAl - STA4 to the mobile network and on at least one of said parameters that characterize the traffic flow of said connection and for sending a response concerning the admission request towards the mobile terminal STAl - STA4.
In the following, by way of example the method according to the invention is described in detail making reference to figs. 1 and 2.
The proposed method for providing channel admission control
(CAC) is implemented in a WLAN system according to the IEEE 802.lie standard and is based on the following assumptions and initial considerations:
• The mobile terminal STA2 requests a new traffic flow by
means of sending a so-called ADDTS request message
specifying the so-called TSPEC element to the access
point API which in turn forwards the ADDTS request mes
sage to the admission controller AC. The TSPEC element
shall include, at least, the following parameters: The
nominal MSDU size, the required mean data rate Rr, the
required minimum PHY rate, and the requested EDCA access
category, which is specified in the so-called TS info
field. Note that the TSPEC element could also specify
the required minimum data rate, which is not essential
to the proposed CAC algorithm, but useful to give to a
mobile terminal the possibility to renegotiate the de
sired QoS level when admission is not possible with the
current QoS requirements. The ADDTS request message and
the TSPEC element are both defined in IEEE 802.lie stan
dard.
• The so-called surplus bandwidth allowance explained be
low is a TSPEC field supposed to be filled by the mobile
terminal STA2. However, due to its extremely complicated
computation, it is very unlikely, although not impossi
ble, that the mobile terminal STA2 can compute it. In
case the surplus bandwidth allowance is not provided by
the mobile terminal STA2, the admission controller AC
shall compute it, using the algorithm defined here:
Draft Supplement to Standard For Telecommunications and
Information Exchange Between Systems - LAN/MAN Specific
Requirements - Part 11: Wireless Medium Access Control
(MAC) and Physical Layer (PHY) specifications: Medium
Access Control (MAC) Enhancements for Quality of Service
(QoS), IEEE Standard 802.lie DO.8.
Alternatively, the surplus bandwidth allowance could be
computed by the access point API and provided to the ad-
mission controller AC. However, this does not seem to be a good solution. In fact, in this case, the access point API must be able to intercept ADDTS request frames, interpret them, compute the surplus bandwidth allowance on the fly, insert it into a new ADDTS request message and forward it to the admission controller AC.
• The metrics of interest are computed over a measurement
window (MW) with the duration of the time period TH de
pending on traffic load and throughput. They are ex
pressed in number of observed successful accesses WA,
i.e. successful transmitted packets, instead of in sec
onds. This number of successful transmitted packets WA
can be measured by the access point API based on the to
tal number of transmitted and received acknowledgements
ACK. This method has the advantage to avoid lack of data
in the case of low traffic and an excessive latency in
the case of high load or load spikes.
In a basic embodiment of the invention, only one class of service, i.e. only one so-called EDCA access category, is defined in the so-called QoS basic service set (QBSS) of the system. Despite such simplification, the description details precisely and completely the chosen approach. The useful parameters will be presented, as well as the metrics measured by the access point API and the ones computed by the admission controller AC. Finally the admission control algorithm itself is detailed.
The following definitions of useful static parameters are introduced to the CAC algorithm proposal:
• The maximum packet drop probability pdrop represents the
maximum probability of any packet to be dropped in the
radio channel, i.e. the probability that the packet
fails to be received within a so-called delay bound.
This value is necessary to compute the surplus bandwidth
allowance. It is assumed to use a pre-configured value,
fixed as a static parameter for the entire basic service set (BSS), i.e. for the admission controller AC. The IEEE 802.lie standard specification proposes to use a value of pdrop = 10"8.
• The target channel utilization Utarget represents the target value of utilization that the admission controller AC wants to achieve and that shall not be exceeded within its controlled QoS basic service set. By definition, the value of Utarget is smaller than 1. The parameter ^target ^s a configurable parameter that can be tuned to adjust the inherent inaccuracy of the CAC algorithm.
In fig. 2, the principle mechanism of the CAC algorithm CALG according to the invention that can be implemented e.g. in the admission controller AC is depicted. The mobile terminal STA2 delivers to the CAC algorithm CALG the parameters STAP that characterize the traffic flow of the connection from the mobile terminal STA2 to the mobile network and that can be comprised e.g. in the TSPEC element. Furthermore, the access point API delivers to the CAC algorithm CALG the parameters APP that characterize the radio channel that shall be used for the connection, like e.g. information about the radio channel utilization. The result OUTPUT of the CAC algorithm CALG comprises e.g. a decision whether or not the traffic flow is admitted and instructions concerning the access of the mobile terminal STA2 to the radio channel. The result OUTPUT is sent from the admission controller AC in which the CAC algorithm CALG is implemented to the mobile terminal STA2.
In the following, the measurements performed by the access point API that lead to the parameters APP that are provided to the admission controller AC as input for the CAC algorithm CALG are described. As already mentioned, the measured metrics of interest are computed over the time period Tw of a measurement window (MW) that is depending on traffic load and throughput and expressed in number of observed successful accesses WA, i.e. successful transmitted packets, instead of in seconds.
• The packet loss probability p2 represents the probability to lose a packet over the wireless medium. It can be derived using both the average packet error probability over the radio channel denoted by pe, which can be easily estimated by the access point API, and the collision probability denoted by pc, which can be estimated using the following formula:
\N-1
'-
In this formula, CWmin denotes the so-called minimum contention window size and N denotes the number of active mobile terminals in a collision domain, i.e. there are N mobile terminals within the transmission range of each other that have data to send.
A packet is lost over the wireless medium whether if it collides or if it suffers from an error on the radio channel. Then, p1 is given by this equation:
P1 = Pe + Pc - PePc
The packet loss probability p1 is used by the admission controller AC to compute the surplus bandwidth allowance described below.
• The measured channel utilization Uaeas provides an indication of the system load level . Uaeas is one of the main metrics to be measured by the access point API . Theoretically it is given by this formula:
TT TR
U -1--L.--S-
" meas _ _ v w
In this formula, TI denotes the channel idle time and TB denotes the channel busy time.
From a practical standpoint, Uaea3 can be derived by the access point API from channel observation, as specified in IEEE 802.11k standard specification. According to the section 11.7.7.6 of the IEEE 802.11k standard specifying the radio resource measurement in 802.11 WLAN networks, the access point API performs the following measure-

merits:
o The CCA (CCA - clear channel assessment) idle histogram, i.e. the histogram of the time intervals during which the PHY clear channel assessment detected an idle medium;
o The CCA busy histogram, i.e. the histogram of the time intervals during which the PHY clear channel assessment detected a busy medium;
o The WAV (NAV = network allocation vector) busy histogram, i.e. the histogram of the time intervals during which the NAV value was set to a positive value.
It must be highlighted that from the CCA histograms it is possible to derive the minimum, maximum and average idle and busy time of the channel during the measurement window. One such histogram, e.g. the CCA busy histogram, might report the probability density function (pdf) denoted b(x) of the number of consecutive busy slots x. From it, the average number of busy slots E(B) can be computed in the following way:
00
E[B]= £x-jb(x)
x = 0
The average number of idle slots E[I] can be computed in a similar way from the CCA idle histogram. The measured channel utilization Umeas can then be computed using the following formula:
The latter formula is a practical computation of the measured channel utilization U
The measured channel utilization is measured/computed during the time period of a measure ment window and provided to the admission control

ler AC to be used in the CAC algorithm during the next measurement window.
To perform admission control, the admission controller AC computes a number of metrics. The input data used for this computation are provided by both the access point API and the mobile terminal STA2 involved in the process. The access point API provides to the admission controller AC the parameters APP that characterize the radio channel that shall be used for the connection like e.g. the measured channel utilization U , the
— ' meas '
packet loss probability pI and the duration of the measurement window denoted Tw. The mobile terminal STA2 provides to the admission controller AC by means of the ADDTS request message the parameters STAP that characterize the traffic flow of the connection from the mobile terminal STA2 to the mobile network, i.e. the description of the to be admitted flow comprised e.g. in the TSPEC element.
The metrics computed by the admission controller AC are listed and described below:
• The number of expected accesses A to the mobile network that the requesting mobile terminal STA2 would have attempted if it had been active during the past measurement window of duration Tw is computed by the admission controller AC by means of the following formula:
In this formula, Rmin denotes the minimum data rate of the new flow and SMSDU denotes the nominal MSDU size.
These two values are specified in the TSPEC element carried by the ADDTS request message, and the duration of the past measurement window Tw is provided by the ac cess point API.
The number of expected accesses A will then be used for the computation of the so-called surplus bandwidth allowance, as will be explained below.

• The medium time parameter TH is defined by the IEEE 802. lie standard and represents the time for the channel utilization of a single flow. It can be derived by the admission controller AC from the flow characteristics specified in the TSPEC element. The medium time Ta
can be obtained in the following way: R }
In this formula,
o jRr is the required mean data rate as specified in the TSPEC element,
o The nominal MSDU size SMSDU is also specified by the
requesting mobile terminal STA2 in the TSPEC element carried in the ADDTS request frame,
o The frame exchange time Tf is given by the transmission time of a frame at the minimum PHY rate of the mobile terminal, plus the so-called short interframe space (SIFS) time interval and the acknowledgement (ACK) transmission time,
o The surplus bandwidth allowance factor SBA specifies the excess allocation of time used to bound the number of dropped packets of the application. This value accounts for retransmissions due to channel errors, collisions, and MAC and PHY overheads that are not specified in the rate information. In other words, it represents the ratio of the actual over-the-air time that the scheduler should allocate for the transmission of MSDU units at the required rates to the time that would be necessary at the minimum PHY rate if there were no packet losses. As such, it must be greater than unity. It can be calculated once we know the maximum packet drop probability, Pdrop' specified by the requesting mobile terminal, the packet loss probability over the wireless medium

Pj, provided by the access point API and the number of expected access attempts A within a specific time frame, which is the time period of a measurement window TH in this case. A procedure to perform this calculation is presented in the Annex H.3.2 of this document:
Draft Supplement to Standard for Telecommunications and Information Exchange Between Systems - LAN/MAN Specific Requirements - Part 11: Wireless Medium Access Control (MAC) and Physical Layer (PHY) specifications: Medium Access Control (MAC) Enhancements for Quality of Service (QoS), IEEE Standard 802.lie DO. 8.
The IEEE 802.lie specification just mentioned also provides full detailed indications about the computation of the medium time parameter TM.
The medium time parameter TM is computed by the admission controller AC and used in the CAC algorithm as estimation of the increase in utilization of the channel requested by the new flow Uadd:
An estimation of the current channel utilization Ucurr of the radio channel is built and maintained by the admission controller AC during the duration of a measurement window Tw. This is done by combining the measured channel utilization Umeas at the end of the previous measurement window as provided by the access point API with the sum of the additional channel utilization Uaad consumed by each newly admitted flow. The duration of the past measurement window Tw is also required for this estimation.
At the beginning of a new measurement window, the admission controller AC sets the value of the current channel utilization f7curr to the value of the measured channel utilization U provided by the access point

API. At that time, the current channel utilization Ucurr reflects precisely the current state of the radio channel . For each during the current measurement window newly accepted flow, the admission controller AC computes the increase in utilization of the channel by the new flow Uadd and updates [7curr in order to take into account the new active flow. The current channel utilization Ucurr can be computed in the following way:
curr add curr
At the end of a measurement window, L7curr is refreshed to the new measured channel utilization Umsas provided by the access point API, and the whole process restarts.
The next section details the proposed CAC algorithm CALG. It massively uses the metrics defined in the previous sections.
The basic principle of the CAC algorithm CALG is the following: The admission controller AC builds and maintains the estimation of the current channel utilization t/curr and uses it to check whether the resources needed for the increase in utilization of the channel by the new flow Uadd are available or not.
The CAC algorithm CALG is executed by the admission controller whenever an ADDTS request message is received, i.e. the reception of such an ADDTS request message acts as a triggering event. The CAC algorithm CALG takes the following metrics as input:
• The current channel utilization Ucurr maintained by the
admission controller AC, as detailed in the previous
section.
• The increase in utilization of the channel requested by
the new flow Uadd computed by the admission controller AC
by means of the TSPEC element, as detailed in the pre
vious section.
Given these input data, the admission controller AC computes the new (theoretical) channel utilization U that it would

have if the new flow was admitted:
Once the new channel utilization Uam is calculated, the admission controller AC can take its admission decision by checking if the new channel utilization Uaew does not exceed a given threshold, the target channel utilization E7taret:. The target channel utilization Utargec represents the target value of utilization that the admission controller AC wants to achieve within its controlled QBSS set.
The admission controller AC can take three possible decision: Accept the new flow and reserve the resources requested, reject it or initiate a renegotiation by proposing a less resource-consuming TSPEC element for the flow.
The admission controller AC will accept the admission of the new flow if:
new target
In that case, the admission controller AC updates the current channel utilization Ucurr in order to take into account the newly accepted flow: U = U
curr new
In the ADDTS response message, the admission controller AC indicates as medium time parameter TM in the TSPEC element the increase in utilization of the channel requested by the new flow Uddd computed for the flow.
The admission controller AC will try to renegotiate the admission of the new flow with the mobile terminal STA2 if:
new target
If there is a value of data rate R'z greater than or equal to the required minimum data rate Rm.n of the new flow specified by the mobile terminal STA2, this value of data rate R'r can replace the mean data rate Rr in the computation of the increase in utilization of the channel requested by the new flow Uadd, i.e. in the computation of the medium time parameter TM, if the following inequation is true:

U' < U
new target
In this formula, U'am denotes the new channel utilization that has been derived by means of using the value of data rate R'r instead of the mean data rate Rz in the computation of the increase in utilization of the channel requested by the new flow
UM.
During the renegotiation process, the current channel utilization Ucurr is not updated with the parameters of the requested flow, and the resources are not reserved.
The admission controller AC will reject the admission of the new flow for any mean data rate Rr greater than or equal to the minimum data rate Rain specified by the by the mobile terminal STA2, if: U > U
new target
Clearly, in case of rejection, the current channel utilization Ucurr is not updated with the parameters of the requested flow.
In a preferred embodiment, the described admission control mechanism is extended to a more realistic case, where multiple classes of traffic/service, i.e. multiple EDCA access categories are defined like e.g. Voice, Video, and Data. The main difference is that now the most important metrics must be measured/computed taking into account the different access categories .
In the following, the modifications requested in the definition of the useful parameters, as well as in the metrics measured by the access point API and the ones computed by the admission controller AC are described. Also, the admission control decision must take into account the existence of the different EDCA access categories.
In the case of multiple classes of service, the static parameters evolve as follows:
• The maximum packet drop probability pdrop rests unchanged w.r.t. its definition given in the basic embodiment, because it is a value that applies to the

whole QBSS set and does not depend on the different classes of service.
The target channel utilization U^'get depends on the
access category ACi. When several classes of service, i.e. EDCA access categories, are defined, the admission controller decides by initial configuration how it wants to share the available bandwidth in its QBSS set among the different access categories. For example, it could choose to reserve 70% of the total available bandwidth to VoIP traffic (VoIP = Voice over Internet Protocol), 20% to video streaming traffic, and the rest (10%) for best-effort data traffic, like e.g. web-browsing. Obviously, the following relation describes the overall target channel utilization
tsaget '
y uACi =u
/ j ^ t arg et t arg et
{ACi}
The measurements performed by the access point API are also impacted by the presence of multiple classes of service:
• The packet loss probability per access category /?/
depends on the collision probability p*a which in turns depends on the minimum contention window size CW*±n . Since the contention window size differs between the different EDCA access categories, the collision probability also depends on the different EDCA access categories. Thus, it must be computed for each
access category, using the corresponding CW*^ i-n the following way:
Therefore, also the packet loss probability differs among the different EDCA access categories:

• The access point API needs to measure the channel
utilization U^s per access category. This can be
easily obtained by using the mechanisms defined in the IEEE 802.11k standard. In fact, the IEEE 802.11k draft standard specifically refers to QoS measurements for different EDCA access categories as described here:
Section 11.7.7.6 of Draft Supplement to Standard for Telecommunication and Information Exchange between Systems - LAN/MAN Specific Requirements - Part 11: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications: Specification for Radio Resource Measurement, IEEE Standard 802.11k DO.8, December 2003.
An alternative solution might be to resort to a Mon-tecarlo approach, by selecting a subset of received/sent packets, e.g. about 15%-20% of the total packets, and perform the estimation on such reduced subset.
In any case, the following relation must be respected by definition:
jL.jU meas U meas {ACi}
Some of the metrics computed by the admission controller AC are impacted by the effect of prioritization among the different classes of service:
• The number of expected accesses A is not impacted by
the presence of several EDCA access categories and
its definition rests unchanged compared with the case
of only one EDCA access category that is described
above.
• The medium time parameter T*Cl and therefore also the

increase in utilization of the channel requested by the new flow U*^ depend on the access category Aci because the surplus bandwidth allowance is computed using the packet loss probability p*Cl that is dependent on the access category ACi. The computation of the medium time parameter T*Cl is straightforward.
• The current channel utilization [/cur is estimated by
the admission controller AC dependent on the access category Aci. The definition that follows is directly derived from the one described above, knowing that a to-be-admitted flow is always mapped on an access category ACi:
yACi 0ACi yACi
curr add curr
To extend the proposed solution for the CAC algorithm CALG to the case where several EDCA access categories are defined, the CAC algorithm CALG must take into account the traffic differentiation. To do so, the following redefined metrics must be
used: The measured channel utilization U^s per access category from the access point API, the target channel utilization U^et per access category, the medium time parameter T*Cl per access category and the current channel utilization U*^r per access category.
The CAC algorithm CALG can be easily derived from the basic case described above using the redefined metrics for the case of multiple access categories. The main idea is that now the admission controller AC performs admission control per EDCA access category, keeping an indication of the current channel
utilization U*^r for each access category ACi.
At the end of each measurement window, the access point API provides to the admission controller AC the measured channel

utilization U^s for each access category ACi.
When an ADDTS request message is received by the admission controller AC, the admission controller AC identifies to which access category ACi the new flow maps by means of the so-called TS Info field of the TSPEC element, and perform exactly the same process as described for the CAC algorithm CALG in case of only one access category ACi, now using the metrics for multiple access categories, i.e. the admission controller AC computes the new (theoretical) channel utilization U^ that it
would have if the new flow were admitted in the requested access category ACi according to the following formula:
TTACi _ r7ACi . rrACi Une« - Ucurr + Uadd
The admission controller AC can now take its decision by checking if the new (theoretical) channel utilization U^ does
not exceed the target channel utilization U^get of the considered access category ACi. Similar to the basic case with only one access category ACi , the admission controller can accept, reject or renegotiate the flow.
The admission controller AC will accept the admission of the new flow if the following inequation is true: UACi< UACi
unew target
In that case, the admission controller AC updates the corresponding current channel utilization Ur in the following way in order to take into account the newly accepted flow:
uACi = UAC1
curr new
In the ADDTS response message sent to the mobile terminal STA2, the admission controller AC indicates as medium time parameter
T*Cl in the TSPEC element the increase in utilization of the channel requested by the new flow U^ computed for the flow.
The admission controller AC will try to renegotiate the ad-

mission of the new flow with the mobile terminal STA2 if: UACi uACi
new target
If there is a value of data rate R'r greater than or equal to the required minimum data rate Rmia of the new flow specified by the mobile terminal STA2, this value of data rate jR'r can replace the mean data rate Rr in the computation of the increase
in utilization of the channel requested by the new flow U*^ ,
i.e. in the computation of the medium time parameter TACl, if the following inequation is true:
T-T i&Ci . T]Aci new utarget
In this formula, U'^ denotes the new channel utilization that
has been derived by means of using the value of data rate R'r instead of the mean data rate Rr in the computation of the increase in utilization of the channel requested by the new flow
UAC1
uadd •
During the renegotiation process, the current channel utilization t/^rr i-s not: updated with the parameters of the requested flow, and the resources are not reserved.
The admission controller AC will reject the admission of the new flow for any mean data rate Rr greater than or equal to the minimum data rate Rmia specified by the by the mobile terminal STA2, if: UACi > UACi
u new ' utarget
Clearly, in case of rejection, the current channel utilization Ucurr is not- updated with the parameters of the requested flow.

Claims
1. A method for providing admission control for the admis
sion of a mobile terminal to a mobile network wherein said
admission control is based on at least one parameter that
characterizes the radio channel that shall be used for a
connection of the mobile terminal to the mobile network
and on at least one parameter that characterizes the traf
fic flow of said connection.
2. A method according to claim 1, wherein at least one of
said at least one parameter that characterizes the radio
channel is used to derive the radio channel utilization of
said radio channel.
3. A method according to claim 1, wherein at least one of
said at least one parameter that characterizes the radio
channel and at least one of said at least one parameter
that characterizes the traffic flow of said connection are
used to derive the additional radio channel utilization
caused by said traffic flow of said connection.
4. A method according to claim 3, wherein said admission
control is based on a comparison of the sum of said radio
channel utilization and said additional radio channel
utilization with a target channel utilization.
5. A method according to claim 1, wherein the at least one
parameter that characterizes the radio channel, the at
least one parameter that characterizes the traffic flow of
said connection, the radio channel utilization and/or the
additional radio channel utilization are given or derived
for different classes of traffic.

6. An admission controller for providing admission control
for the admission of a mobile terminal to a mobile net
work, wherein the admission controller comprises means for
receiving admission requests for the admission of the mo
bile terminal to the mobile network, for receiving parame
ters that characterize the radio channel that shall be
used for a connection of the mobile terminal to the mobile
network, for receiving parameters that characterize the
traffic flow of said connection, for controlling the ad
mission based on at least one of said parameters that
characterize the radio channel that shall be used for a
connection of the mobile terminal to the mobile network
and on at least one of said parameters that characterize
the traffic flow of said connection and for sending a re
sponse concerning the admission request towards the mobile
terminal.
7. An admission controller according to claim 6, wherein
said admission controller is colocated with an access
point of the mobile network.
8. A communication system comprising at least one mobile
terminal and at least one access point (API) wherein said
communication system comprises at least one admission con
troller according to claim 6.