entities are the PCF (Policy Control Function)
and the PEP (Policy Enforcement Point) which is
a functionality of the IP BS manager in the
GGSN. The COPS (Common Open Policy Ser-
vice) protocol is used as the query and response
protocol between the PCF and the PEP. The PCF
is a logical policy decision element that uses
standard IP mechanisms to implement policy in
the IP bearer layer. Hence, it makes decisions
with regard to network based IP policy using pol-
icy rules and communicates them to the IP bearer
manager in the GGSN. Figure 4-1 depicts the
end-to-end QoS management architecture.
The role of the PCF in regard to sessions is first
and foremost to authorise the use of QoS re-
sources to support a service to a specific user.
The PCF may collect the parameters needed in
several ways, e.g. by use of SIP signalling, i.e.
from the SDP information, or the information in
the RSVP FlowSpec. By use of the QoS infor-
mation proper authorisation in the form of IP re-
sources is communicated to the PEP. The PEP
deploys this information to enforce the use of
network recourses. If the user violates the
resource usage, e.g. sends more packets than it
is authorised to, it may drop the exceeding pack-
ets. If the user wishes to renegotiate the resource
usage, new policy information has to be con-
veyed between the PCF and the PEP.
4.3 Bearer Control and Call Control
Integration
This section gives a short overview of how
bearer control and call control can be managed
for a typical VoIP call. It is to be noted that fur-
ther work is needed on these issues.
Voice over IP services over UMTS represent a
major asset for a service provider. The provision-
ing of these services is nevertheless technically
challenging in many regards. One of the issues
that is being addressed in 3GPP is the co-ordina-
tion of the SIP call control procedure with the IP
bearer establishment procedure. These two pro-
cedures need indeed to be co-ordinated to prevent
- Calls to be established prior to resources being
reserved, thus leading to unnecessary call
defects; - Resources being committed before the call is
set-up leading to inefficient use of resources
and users being charged before the called
party picks up; - Theft of service and fraud.
A proposal to co-ordinate the call control with
the bearer control is the two phase commit con-
cept which has been designed in the context of
cable networks. (Distributed Call Signalling
(DCS) specification [24] and Dynamic Quality
of Service specification (DQoS) [25].) The sig-
nalling procedure can be decomposed into four
phases.- Phase 1 establishes the call set-up state at both
ends and authorises the traffic flow between
both ends. - Phase 2 reserves the resources (already autho-
rised in phase 1) for that traffic flow. - Phase 3 reports that the preconditions are met
and ringing is accomplished. - Phase 4 starts when the called party picks up
the phone, the resources are committed.
The resource management framework distin-
guishes between two phases: Reserve and Com-
mit. Reserving resources is the ability to admit
the flow with the QoS requested. The commit-
ment is the effective allocation of specific re-
sources to the flow. Making this distinction
improves system efficiency because it assigns
resources only when necessary, and when the
use of these resources may be charged to a cus-
tomer.UMTS does not support the distinction between
resource reservation and resource commitment
as the bearer set-up procedure includes both the
admission of the bearer and the allocation of
resources to that bearer. Some modifications of
UMTS bearer establishment procedures are thus
needed. The main concern is the radio bearer as
the radio access network holds the scarce and
expensive resources. One way of solving the
problem is to design a radio access bearer reser-
vation procedure. This would trigger the admis-
sion control in the RNC without the radio bearer
being set-up. Later during the commit phase the
radio bearer will be set-up.Phase 1 and phase 3 are SIP message exchanges.
These phases are not dependent on the access
technology itself and therefore they can be app-
lied to UMTS without modifications. Neverthe-
less, they introduce additional signalling over
the air interface as SIP is carried end-to-end.This counterbalances the resource efficiency
benefits of allocating resources only after the
caller has picked up. Additionally, the post pick-
up delay may be large as the resource allocation
procedure in UMTS can be time consuming
(phase 4).