Iur-G Functionality in The GERAN: 1 Introductionasaasasfafafafjygujdasdasd
Iur-G Functionality in The GERAN: 1 Introductionasaasasfafafafjygujdasdasd
Iur-G Functionality in The GERAN: 1 Introductionasaasasfafafafjygujdasdasd
Mnchen, Germany
Tdoc GAHW-000048
Agenda item 6.1.2
Core Network
Iu
Iu
RNS
RNS
Iur
RNC
Iub
Node B
RNC
Iub
Iub
Node B
Node B
Iub
Node B
Tdoc GAHW-000048
Agenda item 6.1.2
Radio
Network
Layer
Control Plane
User Plane
RNSAP
Iur Data
Stream(s)
[Note 1]
Transport
Network
Layer
Transport Network
User Plane
Transport Network
Control Plane
Transport Network
User Plane
ALCAP(Q.2630.1)
[Note 2]
SCCP
STC (Q.2150.1)
MTP3-B
ITUN
MTP3-B
ITUN
SSCF-NNI
SSCF-NNI
SCTP
SSCF-NNI
SSCF-NNI
SCTP
SSCOP
UDP / IP
AAL5
SSCOP
UDP / IP
AAL5
AAL2
ATM
Physical Layer
Tdoc GAHW-000048
Agenda item 6.1.2
It is reiterated that the Iur interface is designed thus in order to enable soft handover for the
UTRAN. It is also pointed out that soft handover is irrelevant for the GERAN, with minimal
advantages.
BTS
Gn
TRX
PCU
BSC
SGSN
Gb/Iu
Iur-g
Gn
BTS
TRX
PCU
BSC
SGSN
Gb/Iu
Tdoc GAHW-000048
Agenda item 6.1.2
The GPRS specification does allow the PCU to be placed at the BSC, but the following
difficulties arise with that implementation:
1. The separation of RLC and MAC functionality is not clearly defined and therefore, the
RLC/MAC layer functionality will have to be at the BSC. Adding of buffering to the
transmission network will result in longer latencies.
2. Since the scheduling of transmissions will occur at the PCU, the resource allocation up to the
BTS will have to be equivalent to setting up a circuit for each air interface time slot. As a
result, there has to be close synchronisation between the BSC and the CCU. The reason for
this is that the RLC/MAC blocks, which are transferred in the transmission network (in these
configurations), are synchronised to the air interface.
3. The signalling overhead on the BTS-to-BSC interface is considerably more. It is estimated
that this increase to be of the order of 25-35%.
4. ARQ will have to occur over the transmission network adding to delay and throughput will
be compromised.
The main advantage with a PCU in the RBS is that both air interface and landline performance
can be optimized, without compromising one for the other.
The UTRAN does not rely on link adaptation or Hybrid ARQ to achieve better throughput
efficiencies. Instead, the reliance is more on fast power control to maintain constant radio
conditions. As a result, the UTRAN did not need to place the RLC/MAC layers at the BTS. The
needs of soft handover dictated that these layers be placed at the RNC.
3.1
The primary advantage of implementing the full Iur functionality within Iur-g is alignment with
UTRAN. However, this alignment does not enhance functionality in any measurable way.
Specification of a protocol stack such as in Figure 2 will ensure that there is little or no change to
the core network procedures towards the GERAN with respect to the UTRAN. For manufacturers
that intend using a common platform for the RNC and the BSC, all interfaces as defined will be
available readily, and it would be possible to use like hardware implementations for the BSC and
the RNC.
3.2
The implementation of Figure 3 will put undue burden on the specification of the Iur interface,
since MAC frames will have to be transported across the BTS/BSC interface prior to forwarding
between the drift BSC and the serving BSC. It is pointed out that this interface is left unspecified
in Release 2000, adding to uncertainties in interoperability of vendor equipment. Moreover,
future standardisation of an Iub-equivalent interface for the GERAN will be impacted adversely.
Some problems that were identified with the placement of the PCU at the BSC in Section 3 are
resurrected. For example, the transmission of MAC frames across the BTS-to-BSC interface and
subsequently over the Iur will require proper dimensioning of the transmission network and close
synchronisation with the CCU on the target cell. It seems irregular that the benefits of moving the
PCU to the BTS will then be countered by implementing user plane transmission over the Iur-g
interface.
Tdoc GAHW-000048
Agenda item 6.1.2
It is also strongly felt that implementation of the RLC/MAC layer at the BSC will adversely
affect possible future evolution of the standard that may require fast resource allocation towards
the MS.
Moreover, there is no need for simultaneous transmission of information towards two base
stations, since TDMA systems have no need for soft handover. Also, current packet forwarding
of GTP-U packets are fully capable of providing low-latency switching of bearers during
handover. It is reiterated that all handovers in the GERAN are going to be hard-handovers.
Thus, there is no need to provide functionality that is irrelevant for the GERAN.
In short, the debate regarding specification of an Iur-g for the user and control plane is one
between strict UTRAN conformance and GERAN radio link optimisation. It is felt that there
should be no compromise on radio link issues. As will be shown in the next section, we
recommend that compromise be stopped at specifying a control plane interface alone.
3.3
Ericsson believe it is of some advantage to have an RNSAP equivalent (subset of the UTRAN
RNSAP) in the GERAN within the BSC. The alternative would be to enhance RANAP
procedures to incorporate functionality corresponding to BSC relocation and handover, an
approach would entail changes to the Iu interface. Changes to the Iu interface should be avoided
unless it is in the common evolution interests of the GERAN and the UTRAN. The specification
of an Iur-g for control plane operation will enable the duration of a handover to be reduced,
although we believe that the handover interruption will mostly be governed by the Physical
channel.
Release 99 of the UMTS specification has a problem with cell update in the absence of an Iur
when the UE has no dedicated channels towards the network. This deficiency is being corrected
in Release 2000, and will be available for use by the GERAN.
In this respect, the benefits of alignment with the UTRAN cannot be ignored. Therefore, Ericsson
have no objection to specification of an Iur-g interface for the Control Plane. It is desirable that
this interface be derived from the Iur interface as specified for the UTRAN.
4 Conclusion
This document has argued against specifying Iur-g functionality for the user plane based on the
following factors:
1. Alignment with UTRAN must not be an overriding factor in specification of the GERAN
internal architecture. Such alignment should be restricted to avoiding or minimising changes
to core network interfaces.
2. The GERAN Release 99 allows implementation of the PCU at various functional
components such as the BTS, BSC. Specification of the Iur-g for the user plane will make it
impossible to consider implementation of the PCU at the BTS. There are compelling reasons
to retain the viability of that option.
Tdoc GAHW-000048
Agenda item 6.1.2
3. The specification of a control plane interface for Iur-g addresses all known issues regarding
efficient relocation of mobile stations across various network domains. Limiting the
specification of the Iur-g to control plane functionality also has advantages for the future
evolution of the GERAN network.
In conclusion, we reiterate that there is no need for specifying user-plane frames across the Iur-g.
It is acceptable that the Iur-g interface allows control plane transport through RNSAP (or a subset
of RNSAP as defined in UTRAN).
References
[1]. 3G.PP, General Packet Radio Service; Service Description; Stage 2 (Release 1999), 3G.PP
23.060, April 2000.
[2]. 3G.PP, UTRAN Iur Interface; General Aspects and Principles, 3G TS 25.420, January
2000.
[3]. 3G.PP, UTRAN Overall Description, 3G.TS 25.401 v3.1.0, January 2000.
[4] Alcatel, Use of Iur and Iu-ps interfaces Handovers, 3G.PP Adhoc on GERAN TDOC 2g00113.
[5] Joint Ericsson/Nokia/Alcatel handover contribution to RAN3, to be submitted.