23-05-2012, 10:11 AM
Neighbor Cell Relation List and Physical Cell Identity Self-Organization in LTE
[attachment=22787]
Abstract
Automation of radio network management is a key
determinant to work reduction for wireless operators. By
replacing time consuming and costly tasks with automatic
mechanisms, operational expenditure can be reduced. In this
paper we present a method for automatic configuration of
locally-unique physical cell identities and neighbor cell relation
lists in 3G Long Term Evolution (LTE). This method makes use
of mobile measurements to update the neighbor cell relation lists
in the cells and to detect local cell identity conflicts, report the
conflicts to the Operation Support Systems (OSS) and resolve
them. The performance of the approach is determined using
simulations of realistically deployed macro networks. Conducted
simulations illustrate the ability of the method to resolve local cell
identity conflicts. In particular, the method is capable of both
accommodating new cells and handling a worst case scenario
where all cells are initiated with the same local cell identities and
where neighbor cell relation lists are empty.
Keywords - Self-organization, Neighbor Cell Relation, Physical
Cell Identity, Autonomic Communication, LTE, WCDMA.
I. INTRODUCTION
The need for even higher data rates, new services, and
improved services has driven the standardization and
development of the 3G Long Term Evolution (LTE). The LTE
concept consists of an evolved radio access network (EUTRAN),
and an evolved packet core (EPC). The Third
Generation Partnership Program (3GPP), has listed a set of
requirements that the LTE concept should fulfill, including
downlink and uplink peak data rates of 100 Mbits/s and 50
Mbits/s, respectively [1][2].
LTE is based on a rather flat architecture compared to 2G
and 3G systems. Each cell is served by an eNodeB (“base
station”), and handovers between cells are handled mainly by
signaling directly between the eNodeBs, and not via any radio
network controller node like in 2G and 3G. The cell broadcasts
an identifying signature, a “fingerprint” (Physical Cell Identity,
PCI), which the mobiles use to identify cells, and as time and
frequency reference. These identifying signatures are not
unique (there are 504 different PCIs in LTE). In addition, we
propose to broadcast a globally unique cell identifier (GID),
which can be detected and reported by the user equipments
(mobiles). Detecting the GID will be more difficult and time
consuming, which in turn implies restrictive use. Since
handover is distributed to the eNodeB it benefits from a
eNodeB managed neighbor cell relation (NCR) list of plausible
handover candidates with connectivity information (e.g. IP
address), as well as a mapping between the PCI and the
globally unique cell identifier, GID. This enables the mobile to
identify cells in measurement reports only by the PCI. Fig. 1
illustrates the concept of neighbors and cell identities. It also
illustrates the O&M interface between the eNodeBs and the
Operations Support System (OSS). For further reading about
LTE we refer to [1].
In parallel with the LTE specification and development, the
Next Generation Mobile Network (NGMN) association of
operators brings forward requirements on management
simplicity and cost efficiency. NGMN has summarized such
requirements on Self-Organizing Networks (SON) in a number
of operator use cases [3]. The vision is that centralized and
decentralized algorithms automate tasks that currently require
significant planning efforts. One use case considers handling of
neighbor cell relations (NCR) lists, which is identified as a
parameter that benefit from self-organization.
Figure 1. OSS monitors neighbor cell lists, and manages PCIs. Some cells
have complete knowledge of theneighbor relationss (cell 1), some cells miss
neighbor relations possibly due to inaccurate models in the planning step (cell
2 and 3) and some cells are newly installed without neighbor info (cell 4).
In 2G and 3G systems NCR lists have been populated using
cell planning tools by means of coverage predictions before the
installation of a base station. Prediction errors, due to
imperfections in map and building data, have forced the
operators to resort to drive/walk tests to completely exhaust the
coverage region and identify all handover regions. This has
proven to be costly and new methods for automatically
deriving NCR lists are required. Furthermore, the LTE
specification includes closed subscriber group (CSG) cells,
1
2
4
3
NCR 1: 2, 3
NCR 3: 1
NCR 4: -
Service
area
Unknown
handover area
NCR 2: 1
OSS
Handover area
O&M
also sometimes denoted Home eNodeBs, which a consumer
may purchase and install in her/his home. This means that
traditional drive/walk test becomes even more difficult.
The second issue in this paper is PCI management. Radio
networks need to handle non-unique local physical identifiers
of cells to support efficient measurement and reporting
procedures. For example, in LTE a mobile is required to
measure the reference signal received power (RSRP) (i.e. the
received power of the signature sequence symbols associated
with a particular PCI) of candidate cells and report to the
serving cell (the cell serving the mobile at the moment). It is
important to detect and resolve local PCI conflicts, i.e. when
two cells in the vicinity of each other uses the same PCI, to
avoid ambiguities in the measurement reports.
The contributions in this paper aid operators in decreasing
their Operating Expenditures (OPEX) by moving NCR list and
PCI management functionality from operators to the system
itself. See further details in [13][17] and coming journal paper.
The remainder of this paper is organized as follows. In
Section II we give an overview of related work. This is
followed by Section III, where the approaches to the problems
stated are described. The performance evaluation is presented
in Section IV, followed by conclusions in Section V.
II. RELATED WORK
PCI conflict resolution corresponds to code planning and
resolution in WCDMA systems. One difference, however, is
that no globally unique cell identity is reported by the mobiles
in WCDMA. There are some papers appearing on code
planning for WCDMA systems, e.g. [4][5], but there is no
literature describing automatic PCI conflict resolution.
In 2G and 3G systems, the mobiles need NCR lists in order
to report candidate cells, but in LTE the mobiles operate
without NCR lists. Instead, it is the eNodeBs that benefit from
the NCR lists. Considering NCR generation, one of early
approaches was formulated for GSM, D-AMPS, and PDC in
[6][7]. In their approach a set of new test cells (frequencies) are
added to the neighbor list of a cell. This enables a mobile to
measure cells currently not on the NCR list of the cell serving
the mobile. The product implementation of the proposed
method is briefly discussed in [8]. In contrast, we propose a
method, which utilizes a feature of LTE, namely that the
mobile detects a new cell and reports to the base station,
making the detection of new cells easier.
In WCDMA, the mobiles are capable of detecting and
reporting cells not listed in the neighbor cell list – detected set
reporting (DSR) [2][9]. Soldani and Ore report results on selfoptimization
of NCR lists for UTRA FDD networks using DSR
measurements [10]. This approach is not directly applicable to
LTE, where it is possible for the mobile to extract the globally
unique cell identifier and report to the eNodeB. Baliosian and
Stadler developed a centralized procedure for creating NCR
lists [11] where each base station intersects the set of mobiles
in its service area with the mobiles in the service area of all
other base stations. Parodi et al. [12] proposed a method for
NCR definition, where the service area of the cells are
approximated and their overlap is computed.
III. APPROACH
With an extensive use of mobile-assisted measurements,
which is already part of the handover procedure, automated
updates of NCR lists and detection and resolution of PCI
conflicts are made possible. For detection of situations where
two cells with overlapping coverage use the same PCI an
additional mechanism is needed. In the following sections,
methods and procedures for NCR management, PCI conflict
detection and PCI conflict resolution are described, giving an
efficient and flexible alternative to drive testing and manual
tuning. The proposed algorithm runs both in eNodeB (NCR
management) and in the Operation Support System, OSS (PCI
conflict resolution). Information will be centralized in the latter
case, but the algorithm acts upon detected PCI conflicts and
uses local information related to the conflicting cell, its
neighbors and neighbors’ neighbors. This means that the
algorithm could be decentralized to the eNodeBs and be based
on signaling between eNodeBs only as is described in [13].
The globally unique cell identifier (GID) in LTE consists of
two parts:
• PLMN Identity: The identity of the Public Land Mobile
Network. Note that a cell may have multiple PLMN
identities.
• CIPL: Unique Cell Identity for a cell within a PLMN [14].
We will assume that whenever a new cell is introduced into the
system it contacts a configuration server in the network. The
configuration server provides the new cell with the GID
identity and an IP address, and other initial parameter values.
Optionally, the configuration server may also provide the cell
with an initial PCI. One way of selecting the initial PCIs, which
allows PCI grouping, could be PCIinitial = CIPL mod A + B
where A is the PCI group size and B is the first PCI in the
corresponding PCI group. PCI grouping can e.g. be used to
ensure that there are no conflicts between macro and micro
cells.
A. NCR Management and PCI Conflict Detection using Handover Measurements
The mobiles continuously measure the RSRP from the
serving cell and candidate cells (cells in the vicinity of the
mobile that might be considered as handover candidates).
A measurement report is typically triggered when the RSRP
from a candidate cell is within a threshold D dB from the
serving cell RSRP.
The measurement report contains information about the
PCI and the corresponding RSRP of the candidate cell. The
serving cell may order the mobile to read the GID (transmitted
on the broadcast channel from each cell) of a cell with a certain
PCI and report that back to the serving cell.
This could be done for example if the PCI is associated
with a cell with handover failures in the past or if a central
node such as the OSS has requested it. In any case, the GID of
a neighboring cell can be obtained with help from a mobile
station upon request from the serving cell. In case the serving
cell decides to set up a relation to the neighboring cell it
contacts the central configuration server in the network and
obtains the IP address (and possibly other connectivity related
information such as encryption and authentication keys).
When a measurement report is received from a mobile it is
handled according to the following scheme:
Is the PCI of the candidate cell already known in the
serving cell (i.e. is the neighbor relation already established)?
Yes: Initiate handover decision procedure.
No: Consider the candidate cell as a NCR list candidate.
Order the UE to report GID. Obtain connectivity
information for the candidate cell and signal to the
candidate cell, directly or through the core network,
about a mutual addition to the NCR lists of the two cells.
Does the candidate cell confirm the NCR list addition?
Yes: Add the candidate cell to the NCR list, and
store relevant information about the cell. Initiate a
handover decision procedure.
No: The candidate cell has detected a PCI conflict
in the NCR list addition procedure. The candidate
cell informs OSS about the PCI. Handover may be
initiated simultaneously through the core network,
using the GID as identifier.
Figure 2. PCI conflict detection based on handover measurements. The
mobile reports a new candidate cell with a PCI already present in the NCR list
of the serving cell.
Two ways of detecting PCI conflicts by using handover
measurements are illustrated in Fig. 2, where cells B and C are
in conflict. Either, cell A is the serving cell, the mobile
approaches cell B and reports it as a candidate cell, and cell A
detects the PCI conflict. Alternatively, cell B is the serving cell
and the mobile approaches cell A. The mobile then reports cell
A as a candidate cell, and cell B adds cell A to its NCR list.
However, when enforcing a mutual neighbor relation, cell A
once again detects the PCI conflict. When a PCI conflict is
detected it is reported to OSS, in both these cases by cell A.
OSS may initiate a network wide re-planning of PCIs when
a conflict is detected. However, the proposal in this paper is to
perform a local adjustment that only involves the PCI update of
one of the conflicting cells based on local information from the
conflicting cell, its neighbors and their neighbors. First, OSS
determines which of the conflicting cells that should change
PCI. It could be at random, or for example be the conflicting
cell i) with the lowest GID, ii) with the shortest NCR list or iii)
that most recently changed PCI that will change the PCI. Only
little differences are reported in [13]. In the following sections,
the cell chosen to change its PCI is referred to as the selected
conflicting cell. The subsequent resolution can be divided into
four steps:
A. Compilation of a set of locally conflicting PCIs by
retrieving the PCIs of the neighbors and neighborsneighbors
to the selected conflicting cell.
B. Determination of the locally non-conflicting PCIs
as all available except the locally conflicting PCIs.
C. Randomization of a new PCI from the locally nonconflicting
PCIs.
D. Information about the PCI update to all cells with
NCR lists containing the selected conflicting cell
B. PCI Detection and Resolution using Transmission Gaps
Another type of PCI conflict is when the candidate cell has
the same PCI as the serving cell. For example, if the mobile is
served by cell C in Fig. 2, and approaches cell B. In that case,
the mobile may not be able to report the weaker cell to the
serving cell because the weaker cell is directly interfered by the
serving cell on the same signature sequence. This comes about
because the mobile cannot differentiate between normal multipath
in the radio channel and the case when the same signature
sequence is transmitted from two different cells. If the two
conflicting cells have at least one neighbor candidate cell in the
vicinity that can detect the conflict, the conflict will eventually
be solved by that cell.
If no such neighbor candidate cells exist or in order to
increase the conflict detection opportunities, the serving cell
could issue a transmission gap at predefined times. During each
such transmission gap the mobile would search for the
signature sequence associated with the serving cell. If the
mobile detects this sequence during a gap it knows that the
signature sequence is transmitted by another cell in the system
(i.e. not the serving cell) and it can inform the serving cell
about the PCI conflict. The start time of the transmission gaps
should be randomized. In order to reduce the risk for
overlapping transmissions gaps between different cells the GID
could be used as a seed in the randomization. The serving cell
can indicate the PCI conflict to OSS, which initiate a PCI
update with the indicating cell as the selected conflicting cell.
Note that OSS does not need to determine the other conflicting
cell.
[attachment=22787]
Abstract
Automation of radio network management is a key
determinant to work reduction for wireless operators. By
replacing time consuming and costly tasks with automatic
mechanisms, operational expenditure can be reduced. In this
paper we present a method for automatic configuration of
locally-unique physical cell identities and neighbor cell relation
lists in 3G Long Term Evolution (LTE). This method makes use
of mobile measurements to update the neighbor cell relation lists
in the cells and to detect local cell identity conflicts, report the
conflicts to the Operation Support Systems (OSS) and resolve
them. The performance of the approach is determined using
simulations of realistically deployed macro networks. Conducted
simulations illustrate the ability of the method to resolve local cell
identity conflicts. In particular, the method is capable of both
accommodating new cells and handling a worst case scenario
where all cells are initiated with the same local cell identities and
where neighbor cell relation lists are empty.
Keywords - Self-organization, Neighbor Cell Relation, Physical
Cell Identity, Autonomic Communication, LTE, WCDMA.
I. INTRODUCTION
The need for even higher data rates, new services, and
improved services has driven the standardization and
development of the 3G Long Term Evolution (LTE). The LTE
concept consists of an evolved radio access network (EUTRAN),
and an evolved packet core (EPC). The Third
Generation Partnership Program (3GPP), has listed a set of
requirements that the LTE concept should fulfill, including
downlink and uplink peak data rates of 100 Mbits/s and 50
Mbits/s, respectively [1][2].
LTE is based on a rather flat architecture compared to 2G
and 3G systems. Each cell is served by an eNodeB (“base
station”), and handovers between cells are handled mainly by
signaling directly between the eNodeBs, and not via any radio
network controller node like in 2G and 3G. The cell broadcasts
an identifying signature, a “fingerprint” (Physical Cell Identity,
PCI), which the mobiles use to identify cells, and as time and
frequency reference. These identifying signatures are not
unique (there are 504 different PCIs in LTE). In addition, we
propose to broadcast a globally unique cell identifier (GID),
which can be detected and reported by the user equipments
(mobiles). Detecting the GID will be more difficult and time
consuming, which in turn implies restrictive use. Since
handover is distributed to the eNodeB it benefits from a
eNodeB managed neighbor cell relation (NCR) list of plausible
handover candidates with connectivity information (e.g. IP
address), as well as a mapping between the PCI and the
globally unique cell identifier, GID. This enables the mobile to
identify cells in measurement reports only by the PCI. Fig. 1
illustrates the concept of neighbors and cell identities. It also
illustrates the O&M interface between the eNodeBs and the
Operations Support System (OSS). For further reading about
LTE we refer to [1].
In parallel with the LTE specification and development, the
Next Generation Mobile Network (NGMN) association of
operators brings forward requirements on management
simplicity and cost efficiency. NGMN has summarized such
requirements on Self-Organizing Networks (SON) in a number
of operator use cases [3]. The vision is that centralized and
decentralized algorithms automate tasks that currently require
significant planning efforts. One use case considers handling of
neighbor cell relations (NCR) lists, which is identified as a
parameter that benefit from self-organization.
Figure 1. OSS monitors neighbor cell lists, and manages PCIs. Some cells
have complete knowledge of theneighbor relationss (cell 1), some cells miss
neighbor relations possibly due to inaccurate models in the planning step (cell
2 and 3) and some cells are newly installed without neighbor info (cell 4).
In 2G and 3G systems NCR lists have been populated using
cell planning tools by means of coverage predictions before the
installation of a base station. Prediction errors, due to
imperfections in map and building data, have forced the
operators to resort to drive/walk tests to completely exhaust the
coverage region and identify all handover regions. This has
proven to be costly and new methods for automatically
deriving NCR lists are required. Furthermore, the LTE
specification includes closed subscriber group (CSG) cells,
1
2
4
3
NCR 1: 2, 3
NCR 3: 1
NCR 4: -
Service
area
Unknown
handover area
NCR 2: 1
OSS
Handover area
O&M
also sometimes denoted Home eNodeBs, which a consumer
may purchase and install in her/his home. This means that
traditional drive/walk test becomes even more difficult.
The second issue in this paper is PCI management. Radio
networks need to handle non-unique local physical identifiers
of cells to support efficient measurement and reporting
procedures. For example, in LTE a mobile is required to
measure the reference signal received power (RSRP) (i.e. the
received power of the signature sequence symbols associated
with a particular PCI) of candidate cells and report to the
serving cell (the cell serving the mobile at the moment). It is
important to detect and resolve local PCI conflicts, i.e. when
two cells in the vicinity of each other uses the same PCI, to
avoid ambiguities in the measurement reports.
The contributions in this paper aid operators in decreasing
their Operating Expenditures (OPEX) by moving NCR list and
PCI management functionality from operators to the system
itself. See further details in [13][17] and coming journal paper.
The remainder of this paper is organized as follows. In
Section II we give an overview of related work. This is
followed by Section III, where the approaches to the problems
stated are described. The performance evaluation is presented
in Section IV, followed by conclusions in Section V.
II. RELATED WORK
PCI conflict resolution corresponds to code planning and
resolution in WCDMA systems. One difference, however, is
that no globally unique cell identity is reported by the mobiles
in WCDMA. There are some papers appearing on code
planning for WCDMA systems, e.g. [4][5], but there is no
literature describing automatic PCI conflict resolution.
In 2G and 3G systems, the mobiles need NCR lists in order
to report candidate cells, but in LTE the mobiles operate
without NCR lists. Instead, it is the eNodeBs that benefit from
the NCR lists. Considering NCR generation, one of early
approaches was formulated for GSM, D-AMPS, and PDC in
[6][7]. In their approach a set of new test cells (frequencies) are
added to the neighbor list of a cell. This enables a mobile to
measure cells currently not on the NCR list of the cell serving
the mobile. The product implementation of the proposed
method is briefly discussed in [8]. In contrast, we propose a
method, which utilizes a feature of LTE, namely that the
mobile detects a new cell and reports to the base station,
making the detection of new cells easier.
In WCDMA, the mobiles are capable of detecting and
reporting cells not listed in the neighbor cell list – detected set
reporting (DSR) [2][9]. Soldani and Ore report results on selfoptimization
of NCR lists for UTRA FDD networks using DSR
measurements [10]. This approach is not directly applicable to
LTE, where it is possible for the mobile to extract the globally
unique cell identifier and report to the eNodeB. Baliosian and
Stadler developed a centralized procedure for creating NCR
lists [11] where each base station intersects the set of mobiles
in its service area with the mobiles in the service area of all
other base stations. Parodi et al. [12] proposed a method for
NCR definition, where the service area of the cells are
approximated and their overlap is computed.
III. APPROACH
With an extensive use of mobile-assisted measurements,
which is already part of the handover procedure, automated
updates of NCR lists and detection and resolution of PCI
conflicts are made possible. For detection of situations where
two cells with overlapping coverage use the same PCI an
additional mechanism is needed. In the following sections,
methods and procedures for NCR management, PCI conflict
detection and PCI conflict resolution are described, giving an
efficient and flexible alternative to drive testing and manual
tuning. The proposed algorithm runs both in eNodeB (NCR
management) and in the Operation Support System, OSS (PCI
conflict resolution). Information will be centralized in the latter
case, but the algorithm acts upon detected PCI conflicts and
uses local information related to the conflicting cell, its
neighbors and neighbors’ neighbors. This means that the
algorithm could be decentralized to the eNodeBs and be based
on signaling between eNodeBs only as is described in [13].
The globally unique cell identifier (GID) in LTE consists of
two parts:
• PLMN Identity: The identity of the Public Land Mobile
Network. Note that a cell may have multiple PLMN
identities.
• CIPL: Unique Cell Identity for a cell within a PLMN [14].
We will assume that whenever a new cell is introduced into the
system it contacts a configuration server in the network. The
configuration server provides the new cell with the GID
identity and an IP address, and other initial parameter values.
Optionally, the configuration server may also provide the cell
with an initial PCI. One way of selecting the initial PCIs, which
allows PCI grouping, could be PCIinitial = CIPL mod A + B
where A is the PCI group size and B is the first PCI in the
corresponding PCI group. PCI grouping can e.g. be used to
ensure that there are no conflicts between macro and micro
cells.
A. NCR Management and PCI Conflict Detection using Handover Measurements
The mobiles continuously measure the RSRP from the
serving cell and candidate cells (cells in the vicinity of the
mobile that might be considered as handover candidates).
A measurement report is typically triggered when the RSRP
from a candidate cell is within a threshold D dB from the
serving cell RSRP.
The measurement report contains information about the
PCI and the corresponding RSRP of the candidate cell. The
serving cell may order the mobile to read the GID (transmitted
on the broadcast channel from each cell) of a cell with a certain
PCI and report that back to the serving cell.
This could be done for example if the PCI is associated
with a cell with handover failures in the past or if a central
node such as the OSS has requested it. In any case, the GID of
a neighboring cell can be obtained with help from a mobile
station upon request from the serving cell. In case the serving
cell decides to set up a relation to the neighboring cell it
contacts the central configuration server in the network and
obtains the IP address (and possibly other connectivity related
information such as encryption and authentication keys).
When a measurement report is received from a mobile it is
handled according to the following scheme:
Is the PCI of the candidate cell already known in the
serving cell (i.e. is the neighbor relation already established)?
Yes: Initiate handover decision procedure.
No: Consider the candidate cell as a NCR list candidate.
Order the UE to report GID. Obtain connectivity
information for the candidate cell and signal to the
candidate cell, directly or through the core network,
about a mutual addition to the NCR lists of the two cells.
Does the candidate cell confirm the NCR list addition?
Yes: Add the candidate cell to the NCR list, and
store relevant information about the cell. Initiate a
handover decision procedure.
No: The candidate cell has detected a PCI conflict
in the NCR list addition procedure. The candidate
cell informs OSS about the PCI. Handover may be
initiated simultaneously through the core network,
using the GID as identifier.
Figure 2. PCI conflict detection based on handover measurements. The
mobile reports a new candidate cell with a PCI already present in the NCR list
of the serving cell.
Two ways of detecting PCI conflicts by using handover
measurements are illustrated in Fig. 2, where cells B and C are
in conflict. Either, cell A is the serving cell, the mobile
approaches cell B and reports it as a candidate cell, and cell A
detects the PCI conflict. Alternatively, cell B is the serving cell
and the mobile approaches cell A. The mobile then reports cell
A as a candidate cell, and cell B adds cell A to its NCR list.
However, when enforcing a mutual neighbor relation, cell A
once again detects the PCI conflict. When a PCI conflict is
detected it is reported to OSS, in both these cases by cell A.
OSS may initiate a network wide re-planning of PCIs when
a conflict is detected. However, the proposal in this paper is to
perform a local adjustment that only involves the PCI update of
one of the conflicting cells based on local information from the
conflicting cell, its neighbors and their neighbors. First, OSS
determines which of the conflicting cells that should change
PCI. It could be at random, or for example be the conflicting
cell i) with the lowest GID, ii) with the shortest NCR list or iii)
that most recently changed PCI that will change the PCI. Only
little differences are reported in [13]. In the following sections,
the cell chosen to change its PCI is referred to as the selected
conflicting cell. The subsequent resolution can be divided into
four steps:
A. Compilation of a set of locally conflicting PCIs by
retrieving the PCIs of the neighbors and neighborsneighbors
to the selected conflicting cell.
B. Determination of the locally non-conflicting PCIs
as all available except the locally conflicting PCIs.
C. Randomization of a new PCI from the locally nonconflicting
PCIs.
D. Information about the PCI update to all cells with
NCR lists containing the selected conflicting cell
B. PCI Detection and Resolution using Transmission Gaps
Another type of PCI conflict is when the candidate cell has
the same PCI as the serving cell. For example, if the mobile is
served by cell C in Fig. 2, and approaches cell B. In that case,
the mobile may not be able to report the weaker cell to the
serving cell because the weaker cell is directly interfered by the
serving cell on the same signature sequence. This comes about
because the mobile cannot differentiate between normal multipath
in the radio channel and the case when the same signature
sequence is transmitted from two different cells. If the two
conflicting cells have at least one neighbor candidate cell in the
vicinity that can detect the conflict, the conflict will eventually
be solved by that cell.
If no such neighbor candidate cells exist or in order to
increase the conflict detection opportunities, the serving cell
could issue a transmission gap at predefined times. During each
such transmission gap the mobile would search for the
signature sequence associated with the serving cell. If the
mobile detects this sequence during a gap it knows that the
signature sequence is transmitted by another cell in the system
(i.e. not the serving cell) and it can inform the serving cell
about the PCI conflict. The start time of the transmission gaps
should be randomized. In order to reduce the risk for
overlapping transmissions gaps between different cells the GID
could be used as a seed in the randomization. The serving cell
can indicate the PCI conflict to OSS, which initiate a PCI
update with the indicating cell as the selected conflicting cell.
Note that OSS does not need to determine the other conflicting
cell.