# US 2022/0286199 A1 · Method for predicting co-channel interference of satellite-to-ground downlink under low-orbit satellite constellation

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2022/0286199 A1
- Document ID: US20220286199A1
- Status: Published application
- Kind: A1
- Application number: 17/637,308
- Filed: 2020-06-12
- Published: 2022-09-08
- Earliest claimed: 2019-08-29
- Assignees: Zhejiang University
- Applicants: Zhejiang University
- Inventors: Chunyi Song; Yuying Xu; Qin Chen; Zhiwei Xu
- Web page: https://patentgrep.com/patent/US20220286199A1

## CPC

- H04B 7/18513: {Transmission in a satellite or space-based system}
- H04B 17/336: Signal-to-interference ratio [SIR] or carrier-to-interference ratio [CIR]
- H04B 17/345: Interference values ({signal-to-interference ratio [SIR] or carrier-to-interference ratio [CIR]} H04B17/336)
- H04B 17/373: Predicting channel quality {or other radio frequency [RF]} parameters
- H04B 17/3913: {Predictive models, e.g. based on neural network models}
- H04W 52/243: {taking into account interferences}
- H04W 52/42: in systems with time, space, frequency or polarisation diversity
- H04B: TRANSMISSION
- H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72)

## Abstract

The present disclosure provides a method for predicting co-channel interference (CCI) of a satellite-to-ground downlink under a low-orbit satellite constellation. The method includes: obtaining CCI values in a first time period and CCI values in a second time period; constructing a CCI model based on the CCI values in the first time period and a dropout wavelet neural network (WNN); updating parameters of the CCI model; inputting the CCI values in the second time period into an updated CCI model to obtain accuracy; determining whether the accuracy is greater than an accuracy threshold; and if the accuracy is greater than the accuracy threshold, reconstructing the CCI model; or if the accuracy is less than or equal to the accuracy threshold, outputting the updated CCI model; and predicting CCI values of each link at a next moment based on the updated CCI model.

## Claims

### Claim 1

A method for predicting co-channel interference (CCI) of a satellite-to-ground downlink under a low-orbit satellite constellation, comprising:
S1: obtaining CCI values in a first time period and CCI values in a second time period;
S2: constructing a CCI model based on the CCI values in the first time period and a dropout wavelet neural network (WNN);
S3: updating parameters of the CCI model;
S4: inputting the CCI values in the second time period into an updated CCI model to obtain accuracy;
S5: determining whether the accuracy is greater than an accuracy threshold; and if the accuracy is greater than the accuracy threshold, going back to “S2”; or if the accuracy is less than or equal to the accuracy threshold, outputting the updated CCI model; and
S6: predicting CCI values of each link at a next moment based on the updated CCI model.

### Claim 2 (depends on 1)

The method for predicting CCI of a satellite-to-ground downlink under a low-orbit satellite constellation according to claim 1, wherein the obtaining CCI values in a first time period specifically comprises:
S1.1: obtaining the CCI values in the first time period by using a satellite; or
S1.2: obtaining the CCI values in the first time period through modeling.

### Claim 3 (depends on 2)

The method for predicting CCI of a satellite-to-ground downlink under a low-orbit satellite constellation according to claim 2, wherein the obtaining the CCI values in the first time period through modeling specifically comprises:
S1.2.1: analyzing a movement relationship between a satellite phased array and a terminal phased array to determine deviation angles, wherein the deviation angles comprise a deviation angle of each interference beam relative to a normal direction of the satellite phased array and a deviation angle of a receive antenna of the terminal phased array in a direction of each interference beam;
S1.2.2: calculating a CCI value caused by each interference beam based on the deviation angles, wherein a specific formula is
CCIq(t)=PT·Gts(φq1(t))·Gre(φq2(t))·(dq(t))−α (1), wherein
CCIq (t) denotes a CCI value caused by a qth interference beam, PT denotes transmit power of a satellite that can generate an interference beam, φq1(t) denotes a deviation angle of the qth interference beam relative to the normal direction of the satellite phased array, φq2(t) denotes a deviation angle of the receive antenna of the terminal phased array in a direction of the qth interference beam, Gts(φq1(t)) denotes an off-axis gain of a transmit antenna of the satellite that can generate an interference beam, Gre(φq2(t)) denotes an off-axis gain of the receive antenna of the terminal phased array, dq(t) denotes a distance from the qth interference beam of the satellite phased array to the terminal phased array, and α denotes a spatial attenuation factor;
S1.2.3: calculating the sum of CCI values caused by all interference beams, and determining CCI values that are caused by all interference beams and that are received by a target beam, wherein a specific formula is
CCI
⁡
(
t
)
=
∑
q
=
1
N
CCI
q
(
t
)
.
(
2
)
wherein
N denotes the total number of interference beams; and
S1.2.4: saving the CCI values that are caused by all interference beams and that are received by the target beam as an array, that is, the CCI values in the first time period are X=(x1, x2, . . . , xn)T, wherein x1=CCI(t0), x2=CCI(t0+Δt), and similarly, xn=CCI(t0+(n−1)Δt);
t0 denotes an initial moment, n denotes the total number of CCI values in the first time period, Δt denotes a sampling time, and CCI(t0+(n−1)Δt) denotes CCI values that are caused by all interference beams and that are received by the target beam at moment t0+(n−1)Δt.

### Claim 4 (depends on 3)

The method for predicting CCI of a satellite-to-ground downlink under a low-orbit satellite constellation according to claim 3, wherein the constructing a CCI model based on the CCI values in the first time period and a dropout WNN specifically comprises:
S2.1: establishing an input layer, a hidden layer, and an output layer based on a dropout three-layer WNN;
S2.2: initializing the total number l of hidden-layer nodes, the total number m of output-layer nodes, a dropout probability p, a connection weight vi, corresponding to a jth hidden-layer node, between an input-layer neuron and a hidden-layer neuron, a connection weight wj, corresponding to a kth output-layer node and the jth hidden-layer node, between the hidden-layer neuron and an output-layer neuron, a hidden-layer threshold ai corresponding to the jth hidden-layer node and an ith input-layer node, and an output-layer threshold bj corresponding to the kth output-layer node and the jth hidden-layer node, and initializing a learning rate η and a neuron excitation function ƒ(x)=e−x2/2·cos(1.75x), using X=(x1, x2, . . . , xn)T as input;
S2.3: determining an input quantity corresponding to each hidden-layer node based on the CCI values in the first time period, wherein a specific formula is
net
j
=
∑
i
=
1
n
(
v
i
·
x
i
+
a
i
)
,
j
=
1
,
2
,
⋯
,
l
,
(
3
)
wherein
netj denotes an input quantity corresponding to the jth hidden-layer node, xi denotes the ith CCI value corresponding to the jth hidden-layer node, and n denotes the total number of CCI values in the first time period;
S2.4: calculating an actual output quantity corresponding to each hidden-layer node, wherein a specific calculation formula is
y
j
=
f
⁡
(
net
j
)
=
e
-
net
j
2
/
2
·
cos
⁡
(
1.75
net
j
)
,
(
4
)
wherein
yj denotes an actual output quantity corresponding to the jth hidden-layer node;
S2.5: calculating an input quantity corresponding to each output-layer node, wherein a specific formula is
{
y
~
j
=
r
j
*
y
j
net
k
=
∑
j
=
1
l
(
w
j
·
y
~
j
+
b
j
)
,
k
=
1
,
2
,
⋯
,
m
,
(
5
)
wherein
{tilde over (y)}j denotes a dropout output quantity, rj denotes a vector of 0 and 1 randomly generated with probability p, and netk denotes an input quantity corresponding to the kth output-layer node; and
S2.6: calculating an actual output quantity corresponding to each output-layer node, wherein the actual output quantity corresponding to the output-layer node is a predicted CCI value, and
a specific formula for calculating the actual output quantity corresponding to each output-layer node is:
o
k
=
f
⁡
(
net
k
)
=
e
-
net
k
2
/
2
·
cos
⁡
(
1.75
net
k
)
,
(
6
)
wherein
ok denotes the actual output quantity corresponding to the kth output-layer node, that is, a kth predicted CCI value.

### Claim 5 (depends on 4)

The method for predicting CCI of a satellite-to-ground downlink under a low-orbit satellite constellation according to claim 4, wherein the updating parameters of the CCI model specifically comprises:
S3.1: calculating an output error corresponding to each CCI value, wherein a specific formula is
ek=dk−ok (7), wherein
ek denotes an output error corresponding to a kth CCI value, and dk denotes the kth CCI value; and
S3.2: updating vi, wj, ai, and bj based on the output error ek, wherein a specific formula is
{
v
i
′
=
v
i
+
η
·
y
j
(
1
-
y
j
)
⁢
x
i
⁢
∑
j
=
1
m
w
j
·
e
k
w
j
′
=
w
j
+
η
·
y
j
·
e
k
a
i
′
=
a
i
+
η
·
y
j
(
1
-
y
j
)
⁢
∑
j
=
1
m
w
j
·
e
k
b
j
′
=
b
j
+
e
k
,
(
8
)
wherein
v′i denotes an updated connection weight, corresponding to the jth hidden-layer node, between the input-layer neuron and the hidden-layer neuron, w′j denotes an updated connection weight, corresponding to the kth output-layer node and the jth hidden-layer node, between the hidden-layer neuron and the output-layer neuron, a′i denotes an updated hidden-layer threshold corresponding to the jth hidden-layer node and the ith CCI value, and b′j denotes an updated output-layer threshold corresponding to the kth output-layer node and the jth hidden-layer node.

### Claim 6 (depends on 1)

The method for predicting CCI of a satellite-to-ground downlink under a low-orbit satellite constellation according to claim 1, wherein the inputting the CCI values in the second time period into an updated CCI model to obtain accuracy specifically comprises:
S4.1: inputting the CCI values in the second time period into the updated CCI model to obtain predicted CCI values in a third time period;
S4.2: obtaining CCI values in the third time period;
S4.3: calculating differences between the CCI values in the third time period and the predicted CCI values in the third time period to obtain a plurality of error values; and
S4.4. determining the accuracy based on the plurality of error values.

### Claim 7 (depends on 6)

The method for predicting CCI of a satellite-to-ground downlink under a low-orbit satellite constellation according to claim 6, wherein the determining the accuracy based on the plurality of error values specifically comprises:
S4.4.1: obtaining the sum of the plurality of error values, wherein the sum of the error values is the accuracy; or
S4.4.2: obtaining the average of the plurality of error values, wherein the average is the accuracy.