Method and apparatus for determining network clusters for wireless backhaul networks
Practical methods and apparatuses are provided for determining network clusters in wireless backhaul networks comprising a plurality of hubs ( 102 ) and Remote Backhaul Modules (RBM) ( 104 ) based on link quality value (LQV) metrics. From an input LQV table of LQV values for each hub-RBM link ( 110 ), the link quality values are first ranked. Clusters are then identified from all the possible links based on the order of the highest link quality value to the lowest link quality value, any constraints on the number of RBMs per cluster, and clustering each RBM only once. Links with strong link quality values are chosen to optimize the LQV metric. LQV based clustering achieves a higher average LQV, e.g., average spectrum efficiency or weighted sum spectrum efficiency, for the entire backhaul network compared to the geographic location based clustering. The method is straightforward to implement and has low computational complexity.
1. A method for determining network clusters in a fixed wireless backhaul network comprising a plurality of hubs and a plurality of Remote Backhaul Modules (RBM) having directional antenna, wherein each cluster comprises a hub serving a number of RBMs (hub-RBM cluster), comprising the steps of:
for N hubs, each of the N hubs serving M RBMs, for a total number of NM RBMs, indexing each of the plurality of N hubs i and each of the plurality of NM RBMs j to define a plurality of all N×NM candidate hub-RBM pairs ij;
for hub-RBM radio links ij between each respective hub-RBM pair ij, and for each candidate location k of each RBM j, obtaining an input link quality value LQV ijk ;
ranking said link quality values LQV ijk of hub-RBM links ij for each candidate location k of each RBM j, and choosing a maximum value LQV ij of these link quality values, where
LQV ij =max k LQV ijk , to obtain a two dimensional input LQV table of values LQV ij of size N×NM;
reshaping the input LQV table into a LQV vector in a column-wise order and ranking link quality values LQV ij to provide a reordered LQV vector, by performing a ranking operation denoted by a permutation π, to reorder the link quality values LQV ij of the LQV vector from the highest to the lowest, and
wherein the inverse permutation of π is denoted by π −1 , which maps the index of the reordered link quality values LQV ij to the index of the reshaped LQV vector,
the LQV vector comprising two dimensions corresponding to a hub index (hub ID) and an RBM index (RBM ID) wherein: for the t th position in the reordered LQV vector, the RBM ID and hub ID are computed as
RBM
ID
(
t
)
=
mod
(
π
-
1
(
t
)
-
1
NM
)
+
1
hub
ID
(
t
)
=
floor
(
π
-
1
(
t
)
-
1
NM
)
+
1
identifying each hub-RBM cluster from the reordered LQV vector, comprising: from the highest to the lowest link quality values, identifying each hub-RBM cluster by excluding each RBM from being clustered more than once and by satisfying a maximum number of RBMs per cluster constraint.
2. The method of claim 1 , wherein the link quality values LQV ijk are weighted and wherein weights are assigned based on traffic conditions of each link.
3. The method of claim 1 , wherein the link quality vector comprises three dimensions, corresponding to said hub index, said RBM index and further comprising a location index, and
wherein the step of ranking link quality values of hub-RBM links further comprises ranking link quality values LQV ijk of multiple candidate site locations k for each RBM j and identifying a respective best location of said candidate locations k of each RBM j.
4. The method of claim 1 , wherein the link quality values LQV ij are approximated as the SINR values for each frequency bin using:
SINR
ij
=
G
ij
[
j
→
i
]
g
ij
P
N
ij
+
∑
l
≠
i
G
lj
[
j
→
i
]
g
li
P
,
where the path loss and shadowing gain between hub i and RBM j in a linear scale is denoted by g ij ; the noise power is denoted by N ij ; the combined transmit and receive antenna gain is denoted by G ij [j→i] assuming the mainlobe of the RBM j's antenna is steered towards hub i; and the transmit power of all hubs is denoted by P.
5. The method of claim 1 , wherein obtaining the link quality values comprises:
a) measuring said link quality values for each of the plurality of hub-RBM links; or
b) computing said link quality values from network parameters; and
wherein link quality values are based on one of:
a data rate of each link; a weighted data rate of each link wherein weights are assigned based on traffic conditions of the links; a SINR value of each link; an average SINR across frequency bins of each link; an average channel gain across frequency bins of each link; and an inverse distance of each link.
6. A method for managing resource allocation for a plurality of radio links in a fixed wireless backhaul network comprising a plurality of hubs and Remote Backhaul Modules (RBM) having directional antenna, wherein each hub serves a number of RBMs (hub-RBM cluster), comprising:
clustering the plurality of hubs and RBMs based on a link quality value (LQV) of each radio link by steps comprising:
for N hubs, each of the N hubs serving M RBMs, for a total number of NM RBMs, indexing each of the plurality of hubs i and each of the plurality of RBMs j to define a plurality of all N×NM candidate hub-RBM pairs ij;
for hub-RBM radio links ij between each respective hub-RBM pair ij, and for each candidate location k of each RBM j, obtaining an input link quality value LQV ijk ;
ranking said link quality values LQV ijk of hub-RBM links ij, for each candidate location k of each RBM j, and choosing a maximum value LQV ij of these link quality values, where LQV ij =max k LQV ijk , to obtain a two dimensional LQV table of size N×NM;
reshaping the input LQV table into a LQV vector in a column-wise order and ranking link quality values LQV ij to provide a reordered LQV vector, by performing a ranking operation denoted by a permutation π, to reorder the link quality values LQV ij of the LQV vector from the highest to the lowest, and
wherein the inverse permutation of π is denoted by π −1 , which maps the index of the reordered link quality values LQV ij to the index of the reshaped LQV vector,
wherein the t th position in the reordered LQV vector, the RBM ID and hub ID are computed as
RBM
ID
(
t
)
=
mod
(
π
-
1
(
t
)
-
1
NM
)
+
1
hub
ID
(
t
)
=
floor
(
π
-
1
(
t
)
-
1
NM
)
+
1
identifying each hub-RBM cluster from the reordered LQV vector, comprising:
from the highest to the lowest link quality links, identifying each hub-RBM cluster by excluding each RBM from being clustered more than once and by satisfying a maximum number of RBMs per cluster constraint; and
scheduling resources to links based on said hub-RBM clusters to provide an increased average link quality value of the wireless backhaul network.
7. A method according to claim 6 further comprising managing power allocation and scheduling for interference mitigation in the network based on said hub-RBM clusters.
8. Apparatus for determining network clusters in a fixed wireless backhaul network comprising a plurality of hubs and a plurality of Remote Backhaul Modules (RBM) having directional antenna, wherein each cluster comprises a hub serving a number of RBMs (hub-RBM cluster), said apparatus comprising a processor and instructions for implementing steps of the method defined in claim 1 .
9. A non-transitory computer readable storage medium storing instructions which, when executed in a processor of a fixed wireless backhaul network comprising a plurality of hubs and a plurality of Remote Backhaul Modules (RBM) having directional antenna, wherein each cluster comprises a hub serving a number of RBMs (hub-RBM cluster), implement steps of the method defined in claim 1 .
10. A method for determining network clusters in a fixed wireless backhaul network comprising a plurality of hubs and a plurality of Remote Backhaul Modules (RBM) having directional antenna, wherein each cluster comprises a hub serving a number of RBMs (hub-RBM cluster), the method comprising the steps of:
for N hubs, each of the N hubs serving M RBMs, for a total number of NM RBMs, indexing each of the plurality of N hubs i and each of the plurality of NM RBMs j to define a plurality of all N×NM candidate hub and RBM pairs ij;
for hub-RBM radio links ij between each respective hub i and RBM j pair, and for each candidate location k of an RBM, obtaining an input link quality value LQV ijk ;
ranking said link quality values LQV ijk of hub-RBM links ij for each candidate location k of an RBM j, and choosing a maximum value LQV ij of these link quality values, where LQV ij =max k LQV ijk , to obtain a two dimensional input LQV table of values LQV ij of size N×NM;
reshaping the input LQV table into a LQV vector in a column-wise order and ranking link quality values to provide a reordered LQV vector;
identifying each hub-RBM cluster from the reordered LQV vector, comprising: from the highest to the lowest link quality values, identifying each hub-RBM cluster by excluding each RBM from being clustered more than once and by satisfying a maximum number of RBMs per cluster constraint; and
wherein the link quality values LQV ij are approximated as the SINR values for each frequency bin using:
SINR
ij
=
G
ij
[
j
→
i
]
g
ij
P
N
ij
+
∑
l
≠
i
G
lj
[
j
→
i
]
g
li
P
,
where by g ij denotes a pathloss and shadowing gain between hub i and RBM j in a linear scale; N ij denotes a noise power; G ij [j→i] denotes a combined transmit and receive antenna gain assuming a mainlobe of the antenna of RBM j is steered towards hub i; and P denotes a transmit power of all hubs.