Sie sind auf Seite 1von 30

Call Admission Control

Schemes in UMTS

Kamala Subramaniam

Advisor
Dr. Arne A. Nilsson
Outline

 Overview of UMTS

 Rationale behind CAC schemes

 Prevalent CAC Schemes

 Conclusions
What is UMTS?

 Universal Mobile Telecommunications Systems

 Member if the 3G (3rd Generation) family

 Developed by ETSI (European Telecommunications


Standards Institute) within the ITU’s (International
Telecommunication Union’s) IMT (International
Mobile Telecommunications ) framework.
Why UMTS?

 Today consumers use the Internet to access


integrated services such as voice, data and
multimedia.
 Next logical step is to provide the same
services with the added feature of mobility
 UMTS provides data up to 2Mbps making
portable videophones a reality
UMTS Quality of Service (QoS)
Classes
3GPP (3rd Generation Partnership Project) defines four classes for UMTS
 Conversation Class: Delay Constrained / Connection Oriented/ Constant Bit Rate
 Streaming Class: Delay Constrained / Connection Oriented / Variable Bit rate
 Interactive Class: Longer Delay Constraints / Connectionless
 Background Class: Best Effort Connectionless Services
Importance of Call Admission
Control (CAC) Schemes in UMTS
 Need to admit calls selectively into the system minimizing call
dropping and call blocking
 Must make efficient use of Network’s Resources
 Must guarantee QoS. Typical QoS parameters maybe:
Blocking Probabilities
Transmission Rates
Delay
Reliability
 Need to multiplex a non-homogeneous mix of traffic within a
limited set of resources and various propagation characteristics.
 Need to incorporate mobility complications and handoff
procedures.
Outline

 Overview of UMTS

 Rationale behind CAC schemes

Prevalent CAC Schemes

 Conclusions
Rationale behind CAC
schemes
A UMTS network showing cellular architecture, where each cell is served by the Node-B and the Radio
Network Controller (RNC) serving a bunch of Node-B’s
CAC Terminology
 New Call: When a mobile user wants to communicate to another, the Mobile Terminal (MT) obtains a
new channel from the Base Station (BS) it hears best. If a channel is available, the BS grants it and a new
call originates

 New Call Blocking Probability (or simply blocking probability): If all channels are busy, the MT is
not granted the channel and the call is blocked.

 Handoff Call: The procedure of moving between cells when a call is in progress is called a “handoff”.
During handoff the MT requests resources from the BS in the cell it is moving to.

 Handoff Call Dropping Probability (or simply dropping probability): When the MT is denied a
channel in the cell it is moving to, the call is dropped.

 Priority: Forced termination of a call in progress is more annoying than blocking of a new calling attempt
from the users point of view. Clearly, handoff calls must be given a higher priority.

 Cell Dwell Time: After entering a cell, the time a MT resides in it.
Outline

 Overview of UMTS

 Rationale behind CAC schemes

 Prevalent CAC Schemes

 Conclusions
CAC Schemes

 Capacity Based Schemes

 Mobility Based Schemes

 Interference Based Schemes

 Adaptive Call Admission Control (ACAC)


Capacity Based Schemes
1. Fixed Guard Channel / Cutoff Priority Scheme.

C = CA + CH;
C: Total Number of Channels
CA: Channels allocated to handle admitted calls (handoff and new)
CH: Guard channels allocated to handle handoff calls
New Call Admitted: if total number of calls (handoff and new) < CA
Handoff Call Admitted: if CA + CH < C

PA= number of on-going calls


DN = number of rejected calls
DH = number of rejected handoff calls

If handoff call request


{ If PA < C, PA = PA + 1, and grant admission
Otherwise, DH = DH + 1, and reject}
If new call request
{ If PA < C, then PA = PA + 1, and grant admission
Otherwise, DN = DN + 1, and reject}
If a call is completed or handoff-ed to another cell
{PA = PA – 1}
Results: Fixed Guard Scheme
Hand off Blocking Probability Call Blocking Probability

policy1.2

1
0.8

0.6
0.6

0.5
0.4

0.3

0.4 0.2
0.2 0.1

0 0

time ( sec ) time ( sec )

Blocking and Dropping Probabilities with no Guard Channels implemented

Hand off Blocking Probability Call Blocking Probability

1 0.7
0.6
0.8
0.5
0.6 0.4

0.4 0.3
0.2
0.2
0.1
0 0

time ( sec ) time ( sec )

Blocking and Dropping Probabilities with 25% Guard Channels


Results: Fixed Guard Scheme
policy
Phb and Pnb vs # Guard channels

1.5
Probability : Phb Handoff Blocking
Blocking

and Pnb 1 Probability


0.5 New Call Blocking
Probability
0
0 20 40 60
# Guard channels

Blocking Probabilities Vs Guard Channels


Capacity Based Schemes
1. Adaptive Fixed Guard Channel Scheme.
 Dropping rate Increases, increase number of guard channels
 Keep Dropping rate below Threshold at all times

τ: Time period for updating measurements


H: handoff calls into cells (both rejected and admitted)
DH: number of rejected handoff calls in the past τ seconds
TH: threshold for handoff call dropping probability

If a handoff call is dropped and


DH/H ≥ αuTH then
CH = min {CH + !, Cmax},
where αu is the threshold chosen as, e.g. 0.9.

If DH/H <= αdTH for N consecutive handoff calls, then


CH = max {CH – 1, Cmin},
where αd is another threshold chosen as e.g., 0.6 and N is an integer chosen as
e.g.,10.
Capacity Based Schemes
1. Fractional Guard Channel Policy

New calls accepted with probability = βi


Handoff Calls accepted with probability = 1
where i is the state of the system
number of available channels
d =
c number of channels in total

“Hot” Vs “Cold”

 Define threshold h >0, e.g., 0.2,0.25 and 0.3


 dc » h, “cold cell”: lots of available channels, βi = 1
number of available channels for new calls = (n - g) – i
 i » H, “hot cell”: lower resources, βi = 0

where i : state of the system


g: number of guard channels
n: total number of channels
H = (1 – h) n - g

New Call Acceptance Probability:


1 i<H

βi = (0.85 + e( H −i) / n )k H ≤ i ≤ n -g
0 i > n -g
 
Results: Fractional Guard
Channel Policy

Blocking Probability of new calls as a function Dropping Probability of handoff calls


of the offered traffic load as a function of the offered traffic load
Rationale: Mobility Based
Schemes
 Users of two types: Low Speed (Pedestrian) users and High Speed (Vehicular) users

 Cell Dwell Times = F (elapsed time in cell, velocity class)

 Pr (call will request a handoff sometime after T) = Lh (t,T) for high-speed ; Ll (t,T) for low-speed
1
 Directional Factor: α ij = ; Ni is the set of neighboring cells to cell i
| Ni |

 Influence curves: α ij Lh (t , T ) for a high speed call


I (i , j , t , T ) = 
α ij Ll (t , T ) for a low speed call

 Total Influence that all ongoing calls exert on cell j:


I ij = ∑ α ij L(t k , T )
k∈S

 At time T, cell j needs to reserve:


Rj = ∑R
i∈N j
ij
Mobility Based Schemes
1. Integral MBCR

1 ~
B ≤ C − R j − Bnew
 used

Pnew =  ~
0 B > C − R j − Bnew
 used
Variations 
 Conservative: Ceiling value of Rj; may waste resources
 Aggressive: Floor value of Rj; may increase dropping rate.

13. Fractional MBCR

1 I
B used ≤ C - R j - B new - 1

 F I
Pnew = 1 − R j B used = C - R j - B new
 I
0
 B used > C - R j - B new

where RjI is the integral part and RjF is the fractional part
Mobility Based Schemes
1. New Call Bounding Scheme

1
 B ≤N &B ≤ C − Bnew
Pnew =  usednew bnd used
0 otherwise

 Hybrid Scheme

1 ~
B ≤ C - R - Bnew & B ≤N
Pnew =  usednew j usednew bnd
0 otherwise

Results: Mobility Based
Schemes

Handoff Call Blocking Probability New Call Blocking Probability


Interference Based Schemes
 Admit user into system only if Interference threshold not passed
 CAC scheme: guarantee dropping probability below threshold at high offered
loads.
Interference Based Schemes
1. Wideband Power-Based Admission Control Strategy

uplink admission criterion: Itotal_old + ∆I > Ithreshold

downlink admission criterion: Ptotal_old + ∆Ptotal > Pthreshold


Itotal
∆I = ∆L Derivation of the load curve
1 − ηUl

N
1
ηUL = (1 + i ).∑ uplink load
W
j =1
1+
( Eb / N o ).Ri .υi
W = Chip Rate,R j = Bit rateof Jth user,υ i = voiceactivityfactor
E b /No = Signal energy per bit / Noise spectralDensity

∆L = 1
W
1+
υ.Eb / N o .R
Interference Based Schemes
1. Throughput Based Admission Control Strategy

Uplink criterion: ηUL + ∆L > ηUL_threshold

N
1
ηUL = (1 + i ).∑ uplink load
W
j =1
1+
( Eb / N o ).Ri .υi
W = Chip Rate,R j = Bit rateof Jth user,υ i = voiceactivityfactor
E b /No = Signal energy per bit / Noise spectralDensity

Downlink criterion: ηDL +∆L > ηDL_threshold

υ j ( Eb / N o ) j
η DL = ∑ j =1 R j .
N
.[(1 − αav )+ iav ]
W
α av : average orthogonality of the cell
i av : the average downlink other - to - own cell interference ration of the cell
Interference Based Schemes
1. CAC Based on Signal to Noise Interference Ratio
uplink algorithm:
Eb Ci / Ri C PG
= = i i
N o ( N + I − Ci ) / W ( N + I i )
PG i : Processing Gain; I i = I - Ci : intereference experienced by the user i.

M-1 users in system, Mth user requesting access, minimum required power for new user is:

~ ( Eb / No ) target , M ( IM + N )
CM =
PGM
I M : interference seen by new user if accepted
I M +N : total power that the BS is receiving

downlink algorithm:
power with which the ith user channel is received at the ith MT:
(1 + N ) i
Ci =
1+( PGi /( Eb / N o ) i )

estimation of needed received power for Mth MT:


~ (1 + N ) M
CM =
1 + ( PGM /( Eb / No) target , M
Results: Interference Based
Schemes

Power-based CAC, downlink, homogeneous traffic distribution: offered traffic vs. accepted traffic and
maximum dropping probability for different values of the ratio Pthr/Pmax.

Interference-based CAC, uplink, homogeneous traffic distribution: offered traffic vs. accepted traffic and
maximum dropping probability for different values of the threshold level.
Adaptive Call Admission Control
(ACAC)
 Limit on acceptable interference threshold ↔ number of users of each service class
in local and neighboring cells

 Obtain tradeoff between the number of voice and data users according to
outage/blocking probability.

 Outage Probability: P[C ≥ W] = δ

(E )
 Acceptable Interference level: γk ≅ b k = S k .W (k = 1,....., L)
I I .R k
o

 Total interference plus noise power received at the BS:


L
I = ∑ N k S k + S out + noW
k =1
ACAC
 Constraint on the number of users:
Lo 6 Lc N kc

∑ N ko Rkoγ ko + ∑
k =1 c =1
∑ ∑α
k =1 i =1
kic Rkcγ kc < W (1 − η )

where η = upper bound on the total received interference (0.1 < η < 0.25)

 Bandwidth utilized by a user ofL class k:


o 6 Lc N kc
C ≅ ∑ N ko Rkoγ ko + ∑ ∑ ∑α kic Rkc γ kc
k =1 c =1 k =1 i =1
Conclusions
 Summarized UMTS CAC schemes from open literature

 CAC schemes classified as capacity based, interference based, mobility based and adaptive

 CAC schemes efficiently utilize system resources in order to:

Guarantee QoS
Minimize Blocking/Dropping Probabilities
Minimize Interference
Provide priority to Handoff Calls
Handle Mobility

 Adaptive CAC’s which may be a combination of the above CAC’s are best for a system design

Das könnte Ihnen auch gefallen