Skip to content
Projects
Groups
Snippets
Help
Loading...
Help
Support
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
O
OpenXG-RAN
Project overview
Project overview
Details
Activity
Releases
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Issues
0
Issues
0
List
Boards
Labels
Milestones
Merge Requests
0
Merge Requests
0
CI / CD
CI / CD
Pipelines
Jobs
Schedules
Analytics
Analytics
CI / CD
Repository
Value Stream
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Create a new issue
Jobs
Commits
Issue Boards
Open sidebar
canghaiwuhen
OpenXG-RAN
Commits
a0ffbdd7
Commit
a0ffbdd7
authored
Dec 15, 2016
by
Xiwen JIANG
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
some code optimisationsi; beamforming weights normalisation
parent
cd2a5347
Changes
10
Hide whitespace changes
Inline
Side-by-side
Showing
10 changed files
with
176 additions
and
48 deletions
+176
-48
openair1/PHY/INIT/lte_init.c
openair1/PHY/INIT/lte_init.c
+9
-2
openair1/PHY/LTE_TRANSPORT/dlsch_modulation.c
openair1/PHY/LTE_TRANSPORT/dlsch_modulation.c
+1
-0
openair1/PHY/MODULATION/beamforming.c
openair1/PHY/MODULATION/beamforming.c
+21
-19
openair1/PHY/MODULATION/compute_bf_weights.c
openair1/PHY/MODULATION/compute_bf_weights.c
+22
-18
openair1/PHY/TOOLS/cmult_vv.c
openair1/PHY/TOOLS/cmult_vv.c
+79
-0
openair1/PHY/TOOLS/defs.h
openair1/PHY/TOOLS/defs.h
+19
-0
openair1/PHY/defs.h
openair1/PHY/defs.h
+1
-0
openair1/SIMULATION/LTE_PHY/dlsim.c
openair1/SIMULATION/LTE_PHY/dlsim.c
+2
-2
openair1/SIMULATION/LTE_PHY/dlsim_tm7.c
openair1/SIMULATION/LTE_PHY/dlsim_tm7.c
+20
-6
targets/RT/USER/lte-softmodem.c
targets/RT/USER/lte-softmodem.c
+2
-1
No files found.
openair1/PHY/INIT/lte_init.c
View file @
a0ffbdd7
...
@@ -1244,6 +1244,10 @@ int phy_init_lte_eNB(PHY_VARS_eNB *eNB,
...
@@ -1244,6 +1244,10 @@ int phy_init_lte_eNB(PHY_VARS_eNB *eNB,
if
(
eNB
->
node_function
!=
NGFI_RCC_IF4p5
)
if
(
eNB
->
node_function
!=
NGFI_RCC_IF4p5
)
common_vars
->
txdata
[
eNB_id
][
i
]
=
(
int32_t
*
)
malloc16_clear
(
fp
->
samples_per_tti
*
10
*
sizeof
(
int32_t
)
);
common_vars
->
txdata
[
eNB_id
][
i
]
=
(
int32_t
*
)
malloc16_clear
(
fp
->
samples_per_tti
*
10
*
sizeof
(
int32_t
)
);
// initialze calibration matrix with idendity matrix
for
(
re
=
0
;
re
<
fp
->
ofdm_symbol_size
;
re
++
)
common_vars
->
tdd_calib_coeffs
[
eNB_id
][
i
][
re
]
=
0x00007fff
;
#ifdef DEBUG_PHY
#ifdef DEBUG_PHY
msg
(
"[openair][LTE_PHY][INIT] lte_common_vars->txdataF_BF[%d][%d] = %p (%d bytes)
\n
"
,
msg
(
"[openair][LTE_PHY][INIT] lte_common_vars->txdataF_BF[%d][%d] = %p (%d bytes)
\n
"
,
eNB_id
,
i
,
common_vars
->
txdataF_BF
[
eNB_id
][
i
],
eNB_id
,
i
,
common_vars
->
txdataF_BF
[
eNB_id
][
i
],
...
@@ -1265,7 +1269,7 @@ int phy_init_lte_eNB(PHY_VARS_eNB *eNB,
...
@@ -1265,7 +1269,7 @@ int phy_init_lte_eNB(PHY_VARS_eNB *eNB,
}
}
else
if
(
i
>
4
)
{
else
if
(
i
>
4
)
{
for
(
re
=
0
;
re
<
fp
->
ofdm_symbol_size
;
re
++
)
for
(
re
=
0
;
re
<
fp
->
ofdm_symbol_size
;
re
++
)
common_vars
->
beam_weights
[
eNB_id
][
i
][
j
][
re
]
=
0x00007fff
/
fp
->
nb_antennas_tx
;
common_vars
->
beam_weights
[
eNB_id
][
i
][
j
][
re
]
=
0x00007fff
/
sqrt
(
fp
->
nb_antennas_tx
);
}
}
#ifdef DEBUG_PHY
#ifdef DEBUG_PHY
msg
(
"[openair][LTE_PHY][INIT] lte_common_vars->beam_weights[%d][%d][%d] = %p (%d bytes)
\n
"
,
msg
(
"[openair][LTE_PHY][INIT] lte_common_vars->beam_weights[%d][%d][%d] = %p (%d bytes)
\n
"
,
...
@@ -1314,7 +1318,10 @@ int phy_init_lte_eNB(PHY_VARS_eNB *eNB,
...
@@ -1314,7 +1318,10 @@ int phy_init_lte_eNB(PHY_VARS_eNB *eNB,
}
}
}
//eNB_id
}
//eNB_id
/* Create thread pool */
// Read TDD calibration coefficients
read_calibration_matrix
(
eNB
->
common_vars
.
tdd_calib_coeffs
[
0
],
"PROJECTS/TDDREC/results/calibF.m"
,
fp
);
// Create thread pool
eNB
->
pool
=
new_thread_pool
(
do_OFDM_mod_thread
,
eNB
);
eNB
->
pool
=
new_thread_pool
(
do_OFDM_mod_thread
,
eNB
);
sleep
(
1
);
sleep
(
1
);
...
...
openair1/PHY/LTE_TRANSPORT/dlsch_modulation.c
View file @
a0ffbdd7
...
@@ -1984,6 +1984,7 @@ int dlsch_modulation(PHY_VARS_eNB* phy_vars_eNB,
...
@@ -1984,6 +1984,7 @@ int dlsch_modulation(PHY_VARS_eNB* phy_vars_eNB,
}
}
// mapping ue specific beamforming weights from UE specified DLSCH structure to common space
// mapping ue specific beamforming weights from UE specified DLSCH structure to common space
// TO DO: this doesn't have to be done here since we do it once in one frame
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antennas_tx
;
aa
++
){
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antennas_tx
;
aa
++
){
memcpy
(
phy_vars_eNB
->
common_vars
.
beam_weights
[
0
][
5
][
aa
],
dlsch0
->
ue_spec_bf_weights
[
0
][
aa
],
frame_parms
->
ofdm_symbol_size
*
sizeof
(
int32_t
));
memcpy
(
phy_vars_eNB
->
common_vars
.
beam_weights
[
0
][
5
][
aa
],
dlsch0
->
ue_spec_bf_weights
[
0
][
aa
],
frame_parms
->
ofdm_symbol_size
*
sizeof
(
int32_t
));
}
}
...
...
openair1/PHY/MODULATION/beamforming.c
View file @
a0ffbdd7
...
@@ -64,29 +64,31 @@ int beam_precoding(int32_t **txdataF,
...
@@ -64,29 +64,31 @@ int beam_precoding(int32_t **txdataF,
memset
(
txdataF_BF
[
aa
],
0
,
sizeof
(
int32_t
)
*
(
frame_parms
->
ofdm_symbol_size
));
memset
(
txdataF_BF
[
aa
],
0
,
sizeof
(
int32_t
)
*
(
frame_parms
->
ofdm_symbol_size
));
for
(
p
=
0
;
p
<
14
;
p
++
)
{
for
(
p
=
0
;
p
<
14
;
p
++
)
{
/*
if
(
p
==
0
||
p
==
1
||
p
==
5
)
{
if (p==0 || p==1 || p==5)
//multadd_cpx_vector((int16_t*)txdataF[p],(int16_t*)beam_weights[p][aa], (int16_t*)txdataF_BF[aa], 0, frame_parms->ofdm_symbol_size, 15);
multadd_cpx_vector((int16_t*)txdataF[p],(int16_t*)beam_weights[p][aa], (int16_t*)txdataF_BF[aa], 0, frame_parms->ofdm_symbol_size, 15);
*/
for
(
re
=
0
;
re
<
frame_parms
->
ofdm_symbol_size
;
re
++
)
{
for
(
re
=
0
;
re
<
frame_parms
->
ofdm_symbol_size
;
re
++
)
{
if
(
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
]
!=
0
)
{
if
((
p
==
0
||
p
==
1
||
p
==
5
)
&&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
]
!=
0
)
{
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
0
]
+=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
0
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
0
])
>>
15
);
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
0
]
+=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
0
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
0
])
>>
15
);
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
0
]
-=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
1
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
1
])
>>
15
);
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
0
]
-=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
1
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
1
])
>>
15
);
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
1
]
+=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
0
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
1
])
>>
15
);
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
1
]
+=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
0
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
1
])
>>
15
);
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
1
]
+=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
1
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
0
])
>>
15
);
((
int16_t
*
)
&
txdataF_BF
[
aa
][
re
])[
1
]
+=
(
int16_t
)((((
int16_t
*
)
&
txdataF
[
p
][
slot_offset_F
+
symbol
*
frame_parms
->
ofdm_symbol_size
+
re
])[
1
]
*
((
int16_t
*
)
&
beam_weights
[
p
][
aa
][
re
])[
0
])
>>
15
);
/*
/*
printf("beamforming.c:txdataF[%d][%d]=%d+j%d, beam_weights[%d][%d][%d]=%d+j%d,txdata_BF[%d][%d]=%d+j%d\n",
printf("beamforming.c:txdataF[%d][%d]=%d+j%d, beam_weights[%d][%d][%d]=%d+j%d,txdata_BF[%d][%d]=%d+j%d\n",
p,slot_offset_F+symbol*frame_parms->ofdm_symbol_size+re,
p,slot_offset_F+symbol*frame_parms->ofdm_symbol_size+re,
((int16_t*)&txdataF[p][slot_offset_F+symbol*frame_parms->ofdm_symbol_size+re])[0],
((int16_t*)&txdataF[p][slot_offset_F+symbol*frame_parms->ofdm_symbol_size+re])[0],
((int16_t*)&txdataF[p][slot_offset_F+symbol*frame_parms->ofdm_symbol_size+re])[1],
((int16_t*)&txdataF[p][slot_offset_F+symbol*frame_parms->ofdm_symbol_size+re])[1],
p,aa,re,
p,aa,re,
((int16_t*)&beam_weights[p][aa][re])[0],((int16_t*)&beam_weights[p][aa][re])[1],
((int16_t*)&beam_weights[p][aa][re])[0],((int16_t*)&beam_weights[p][aa][re])[1],
aa,re,
aa,re,
((int16_t*)&txdataF_BF[aa][re])[0],
((int16_t*)&txdataF_BF[aa][re])[0],
((int16_t*)&txdataF_BF[aa][re])[1]);
((int16_t*)&txdataF_BF[aa][re])[1]);
*/
*/
}
}
}
}
}
}
}
return
0
;
return
0
;
}
}
openair1/PHY/MODULATION/compute_bf_weights.c
View file @
a0ffbdd7
...
@@ -7,13 +7,13 @@ int read_calibration_matrix(int32_t **tdd_calib_coeffs, char *calibF_fname, LTE_
...
@@ -7,13 +7,13 @@ int read_calibration_matrix(int32_t **tdd_calib_coeffs, char *calibF_fname, LTE_
FILE
*
calibF_fd
;
FILE
*
calibF_fd
;
int
aa
,
re
,
calibF_e
;
int
aa
,
re
,
calibF_e
;
printf
(
"Number of antennas = %d
\n
"
,
frame_parms
->
nb_antennas_tx
)
;
//
printf("Number of antennas = %d\n", frame_parms->nb_antennas_tx) ;
printf
(
"OFDM symbol size = %d
\n
"
,
frame_parms
->
ofdm_symbol_size
)
;
//
printf("OFDM symbol size = %d\n", frame_parms->ofdm_symbol_size) ;
calibF_fd
=
fopen
(
calibF_fname
,
"r"
)
;
calibF_fd
=
fopen
(
calibF_fname
,
"r"
)
;
if
(
calibF_fd
==
NULL
)
{
if
(
calibF_fd
==
NULL
)
{
printf
(
"ERR: %s not found, running with defaults
\n
"
,
calibF_fname
);
printf
(
"ERR: %s not found, running with defaults
\n
"
,
calibF_fname
);
return
1
;
return
(
1
)
;
}
}
printf
(
"Loading Calibration matrix from %s
\n
"
,
calibF_fname
);
printf
(
"Loading Calibration matrix from %s
\n
"
,
calibF_fname
);
...
@@ -21,33 +21,37 @@ int read_calibration_matrix(int32_t **tdd_calib_coeffs, char *calibF_fname, LTE_
...
@@ -21,33 +21,37 @@ int read_calibration_matrix(int32_t **tdd_calib_coeffs, char *calibF_fname, LTE_
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antennas_tx
;
aa
++
)
{
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antennas_tx
;
aa
++
)
{
for
(
re
=
0
;
re
<
frame_parms
->
ofdm_symbol_size
;
re
++
)
{
for
(
re
=
0
;
re
<
frame_parms
->
ofdm_symbol_size
;
re
++
)
{
fscanf
(
calibF_fd
,
"%d"
,
&
calibF_e
)
;
fscanf
(
calibF_fd
,
"%d"
,
&
calibF_e
)
;
printf
(
"aa=%d, re=%d, tdd_calib[0]=%d
\n
"
,
aa
,
re
,
calibF_e
);
//
printf("aa=%d, re=%d, tdd_calib[0]=%d\n", aa, re, calibF_e);
((
int16_t
*
)(
&
tdd_calib_coeffs
[
aa
][
re
]))[
0
]
=
calibF_e
;
((
int16_t
*
)(
&
tdd_calib_coeffs
[
aa
][
re
]))[
0
]
=
calibF_e
;
fscanf
(
calibF_fd
,
"%d"
,
&
calibF_e
)
;
fscanf
(
calibF_fd
,
"%d"
,
&
calibF_e
)
;
printf
(
"aa=%d, re=%d, tdd_calib[1]=%d
\n
"
,
aa
,
re
,
calibF_e
);
//
printf("aa=%d, re=%d, tdd_calib[1]=%d\n", aa, re, calibF_e);
((
int16_t
*
)(
&
tdd_calib_coeffs
[
aa
][
re
]))[
1
]
=
calibF_e
;
((
int16_t
*
)(
&
tdd_calib_coeffs
[
aa
][
re
]))[
1
]
=
calibF_e
;
//printf("aa=%d, re=%d, tdd_calib=%d+i%d\n", aa, re, (int16_t*)(&tdd_calib_coeffs[aa][re])[0],(int16_t*)(&tdd_calib_coeffs[aa][re])[1]);
//printf("aa=%d, re=%d, tdd_calib=%d+i%d\n", aa, re, (int16_t*)(&tdd_calib_coeffs[aa][re])[0],(int16_t*)(&tdd_calib_coeffs[aa][re])[1]);
}
}
}
}
}
}
void
estimate_DLCSI_from_ULCSI
(
int32_t
**
calib_dl_ch_estimates
,
int32_t
**
ul_ch_estimates
,
int32_t
**
tdd_calib_coeffs
,
LTE_DL_FRAME_PARMS
*
frame_parms
)
{
int
estimate_DLCSI_from_ULCSI
(
int32_t
**
calib_dl_ch_estimates
,
int32_t
**
ul_ch_estimates
,
int32_t
**
tdd_calib_coeffs
,
LTE_DL_FRAME_PARMS
*
frame_parms
)
{
int
aa
,
re
;
int
aa
,
re
;
if
(
frame_parms
->
nb_antennas_tx
!=
frame_parms
->
nb_antennas_rx
)
{
printf
(
"ERR : Tx and Rx antennas should be the same for TDD calibration!"
);
return
(
1
);
}
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antennas_tx
;
aa
++
)
{
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antennas_tx
;
aa
++
)
{
/*multadd_cpx_vector((int16_t*)(&tdd_calib_coeffs[aa][0]),(int16_t*)(&ul_ch_estimates[aa][0]),(int16_t*)(&calib_dl_ch_estimates[aa][0]),1,nb_freq,15);*/
multadd_cpx_vector
((
int16_t
*
)(
&
tdd_calib_coeffs
[
aa
][
0
]),(
int16_t
*
)(
&
ul_ch_estimates
[
aa
][
0
]),(
int16_t
*
)(
&
calib_dl_ch_estimates
[
aa
][
0
]),
1
,
frame_parms
->
ofdm_symbol_size
,
15
);
for
(
re
=
0
;
re
<
frame_parms
->
ofdm_symbol_size
;
re
++
)
{
/*
for (re=0; re<frame_parms->ofdm_symbol_size; re++) {
((int16_t*)(&calib_dl_ch_estimates[aa][re]))[0] += (((int16_t*)(&tdd_calib_coeffs[aa][re]))[0]*((int16_t*)(&ul_ch_estimates[aa][re]))[0])>>15;
((int16_t*)(&calib_dl_ch_estimates[aa][re]))[0] += (((int16_t*)(&tdd_calib_coeffs[aa][re]))[0]*((int16_t*)(&ul_ch_estimates[aa][re]))[0])>>15;
((int16_t*)(&calib_dl_ch_estimates[aa][re]))[0] -= (((int16_t*)(&tdd_calib_coeffs[aa][re]))[1]*((int16_t*)(&ul_ch_estimates[aa][re]))[1])>>15;
((int16_t*)(&calib_dl_ch_estimates[aa][re]))[0] -= (((int16_t*)(&tdd_calib_coeffs[aa][re]))[1]*((int16_t*)(&ul_ch_estimates[aa][re]))[1])>>15;
((int16_t*)(&calib_dl_ch_estimates[aa][re]))[1] += (((int16_t*)(&tdd_calib_coeffs[aa][re]))[0]*((int16_t*)(&ul_ch_estimates[aa][re]))[1])>>15;
((int16_t*)(&calib_dl_ch_estimates[aa][re]))[1] += (((int16_t*)(&tdd_calib_coeffs[aa][re]))[0]*((int16_t*)(&ul_ch_estimates[aa][re]))[1])>>15;
((
int16_t
*
)(
&
calib_dl_ch_estimates
[
aa
][
re
]))[
1
]
+=
(((
int16_t
*
)(
&
tdd_calib_coeffs
[
aa
][
re
]))[
1
]
*
((
int16_t
*
)(
&
ul_ch_estimates
[
aa
][
re
]))[
0
])
>>
15
;
((int16_t*)(&calib_dl_ch_estimates[aa][re]))[1] += (((int16_t*)(&tdd_calib_coeffs[aa][re]))[1]*((int16_t*)(&ul_ch_estimates[aa][re]))[0])>>15;
*/
/*printf("aa=%d, re=%d tdd_calib_coeffs= (%d, %d), ul_ch_estimates = (%d, %d), calib_dl_ch_estimates = (%d, %d)\n",
/*printf("aa=%d, re=%d tdd_calib_coeffs= (%d, %d), ul_ch_estimates = (%d, %d), calib_dl_ch_estimates = (%d, %d)\n",
aa, re,
aa, re,
((int16_t*)&tdd_calib_coeffs[aa][re])[0], ((int16_t*)&tdd_calib_coeffs[aa][re])[1],
((int16_t*)&tdd_calib_coeffs[aa][re])[0], ((int16_t*)&tdd_calib_coeffs[aa][re])[1],
((int16_t*)&ul_ch_estimates[aa][re])[0], ((int16_t*)&ul_ch_estimates[aa][re])[1],
((int16_t*)&ul_ch_estimates[aa][re])[0], ((int16_t*)&ul_ch_estimates[aa][re])[1],
((int16_t*)&calib_dl_ch_estimates[aa][re])[0], ((int16_t*)&calib_dl_ch_estimates[aa][re])[1]);*/
((int16_t*)&calib_dl_ch_estimates[aa][re])[0], ((int16_t*)&calib_dl_ch_estimates[aa][re])[1]);*/
}
//
}
}
}
}
}
...
@@ -89,31 +93,31 @@ void main() {
...
@@ -89,31 +93,31 @@ void main() {
printf("Test Compute BF weights.\n");
printf("Test Compute BF weights.\n");
int32_t **tdd_calib_coeffs, **calib_dl_ch_estimates, **ul_ch_estimates, **beam_weights;
int32_t **tdd_calib_coeffs, **calib_dl_ch_estimates, **ul_ch_estimates, **beam_weights;
int nb_ant, nb_freq, aa, re;
int nb_ant, nb_freq, aa, re;
char calibF_fname[] = "calibF.m";
char calibF_fname[] = "calibF.m";
char BF_fname[] = "BF_weights.m";
char BF_fname[] = "BF_weights.m";
FILE *BF_weights_fd;
FILE *BF_weights_fd;
nb_ant = 8;
nb_ant = 8;
nb_freq = 300;
nb_freq = 300;
// memory allocation
// memory allocation
tdd_calib_coeffs = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
tdd_calib_coeffs = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
calib_dl_ch_estimates = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
calib_dl_ch_estimates = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
ul_ch_estimates = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
ul_ch_estimates = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
beam_weights = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
beam_weights = (int32_t **)malloc(nb_ant*sizeof(int32_t *));
for (aa=0; aa<nb_ant; aa++) {
for (aa=0; aa<nb_ant; aa++) {
tdd_calib_coeffs[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
tdd_calib_coeffs[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
calib_dl_ch_estimates[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
calib_dl_ch_estimates[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
ul_ch_estimates[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
ul_ch_estimates[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
beam_weights[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
beam_weights[aa] = (int32_t *)malloc(nb_freq*sizeof(int32_t));
}
}
// ul channel estimation initilisation
// ul channel estimation initilisation
for (aa=0; aa<nb_ant; aa++)
for (aa=0; aa<nb_ant; aa++)
for (re=0; re<nb_freq; re++)
for (re=0; re<nb_freq; re++)
ul_ch_estimates[aa][re] = 0x7fff7fff;
ul_ch_estimates[aa][re] = 0x7fff7fff;
// calibration coefficients loading
// calibration coefficients loading
read_calibration_matrix(calibF_fname, nb_ant, nb_freq, tdd_calib_coeffs);
read_calibration_matrix(calibF_fname, nb_ant, nb_freq, tdd_calib_coeffs);
...
...
openair1/PHY/TOOLS/cmult_vv.c
View file @
a0ffbdd7
...
@@ -27,6 +27,7 @@
...
@@ -27,6 +27,7 @@
#if defined(__x86_64__) || defined(__i386__)
#if defined(__x86_64__) || defined(__i386__)
int16_t
conjug
[
8
]
__attribute__
((
aligned
(
16
)))
=
{
-
1
,
1
,
-
1
,
1
,
-
1
,
1
,
-
1
,
1
}
;
int16_t
conjug
[
8
]
__attribute__
((
aligned
(
16
)))
=
{
-
1
,
1
,
-
1
,
1
,
-
1
,
1
,
-
1
,
1
}
;
int16_t
conjug2
[
8
]
__attribute__
((
aligned
(
16
)))
=
{
1
,
-
1
,
1
,
-
1
,
1
,
-
1
,
1
,
-
1
}
;
#define simd_q15_t __m128i
#define simd_q15_t __m128i
#define simdshort_q15_t __m64
#define simdshort_q15_t __m64
#elif defined(__arm__)
#elif defined(__arm__)
...
@@ -124,3 +125,81 @@ int mult_cpx_conj_vector(int16_t *x1,
...
@@ -124,3 +125,81 @@ int mult_cpx_conj_vector(int16_t *x1,
return
(
0
);
return
(
0
);
}
}
int
multadd_cpx_vector
(
int16_t
*
x1
,
int16_t
*
x2
,
int16_t
*
y
,
uint8_t
zero_flag
,
uint32_t
N
,
int
output_shift
)
{
// Multiply elementwise the complex conjugate of x1 with x2.
// x1 - input 1 in the format |Re0 Im0 Re1 Im1|,......,|Re(N-2) Im(N-2) Re(N-1) Im(N-1)|
// We assume x1 with a dinamic of 15 bit maximum
//
// x2 - input 2 in the format |Re0 Im0 Re1 Im1|,......,|Re(N-2) Im(N-2) Re(N-1) Im(N-1)|
// We assume x2 with a dinamic of 14 bit maximum
///
// y - output in the format |Re0 Im0 Re1 Im1|,......,|Re(N-2) Im(N-2) Re(N-1) Im(N-1)|
//
// zero_flag - Set output (y) to zero prior to disable accumulation
//
// N - the size f the vectors (this function does N cpx mpy. WARNING: N>=4;
//
// output_shift - shift to be applied to generate output
uint32_t
i
;
// loop counter
simd_q15_t
*
x1_128
;
simd_q15_t
*
x2_128
;
simd_q15_t
*
y_128
;
#if defined(__x86_64__) || defined(__i386__)
simd_q15_t
tmp_re
,
tmp_im
;
simd_q15_t
tmpy0
,
tmpy1
;
#elif defined(__arm__)
int32x4_t
tmp_re
,
tmp_im
;
int32x4_t
tmp_re1
,
tmp_im1
;
int16x4x2_t
tmpy
;
int32x4_t
shift
=
vdupq_n_s32
(
-
output_shift
);
#endif
x1_128
=
(
simd_q15_t
*
)
&
x1
[
0
];
x2_128
=
(
simd_q15_t
*
)
&
x2
[
0
];
y_128
=
(
simd_q15_t
*
)
&
y
[
0
];
// we compute 4 cpx multiply for each loop
for
(
i
=
0
;
i
<
(
N
>>
2
);
i
++
)
{
#if defined(__x86_64__) || defined(__i386__)
tmp_re
=
_mm_sign_epi16
(
*
x1_128
,
*
(
__m128i
*
)
&
conjug2
[
0
]);
tmp_re
=
_mm_madd_epi16
(
tmp_re
,
*
x2_128
);
tmp_im
=
_mm_shufflelo_epi16
(
*
x1_128
,
_MM_SHUFFLE
(
2
,
3
,
0
,
1
));
tmp_im
=
_mm_shufflehi_epi16
(
tmp_im
,
_MM_SHUFFLE
(
2
,
3
,
0
,
1
));
tmp_im
=
_mm_madd_epi16
(
tmp_im
,
*
x2_128
);
tmp_re
=
_mm_srai_epi32
(
tmp_re
,
output_shift
);
tmp_im
=
_mm_srai_epi32
(
tmp_im
,
output_shift
);
tmpy0
=
_mm_unpacklo_epi32
(
tmp_re
,
tmp_im
);
//print_ints("unpack lo:",&tmpy0[i]);
tmpy1
=
_mm_unpackhi_epi32
(
tmp_re
,
tmp_im
);
//print_ints("unpack hi:",&tmpy1[i]);
if
(
zero_flag
==
1
)
*
y_128
=
_mm_packs_epi32
(
tmpy0
,
tmpy1
);
else
*
y_128
=
_mm_adds_epi16
(
*
y_128
,
_mm_packs_epi32
(
tmpy0
,
tmpy1
));
//print_shorts("*y_128:",&y_128[i]);
#elif defined(__arm__)
msg
(
"mult_cpx_vector not implemented for __arm__"
);
#endif
x1_128
++
;
x2_128
++
;
y_128
++
;
}
_mm_empty
();
_m_empty
();
return
(
0
);
}
openair1/PHY/TOOLS/defs.h
View file @
a0ffbdd7
...
@@ -124,6 +124,25 @@ int mult_cpx_conj_vector(int16_t *x1,
...
@@ -124,6 +124,25 @@ int mult_cpx_conj_vector(int16_t *x1,
uint32_t
N
,
uint32_t
N
,
int
output_shift
);
int
output_shift
);
/*!
Element-wise multiplication and accumulation of two complex vectors x1 and x2.
@param x1 - input 1 in the format |Re0 Im0 Re1 Im1|,......,|Re(N-2) Im(N-2) Re(N-1) Im(N-1)|
We assume x1 with a dinamic of 15 bit maximum
@param x2 - input 2 in the format |Re0 Im0 Re1 Im1|,......,|Re(N-2) Im(N-2) Re(N-1) Im(N-1)|
We assume x2 with a dinamic of 14 bit maximum
@param y - output in the format |Re0 Im0 Re1 Im1|,......,|Re(N-2) Im(N-2) Re(N-1) Im(N-1)|
@param zero_flag Set output (y) to zero prior to accumulation
@param N - the size f the vectors (this function does N cpx mpy. WARNING: N>=4;
@param output_shift - shift to be applied to generate output
*/
int
multiadd_cpx_vector
(
int16_t
*
x1
,
int16_t
*
x2
,
int16_t
*
y
,
uint32_t
N
,
int
output_shift
);
// lte_dfts.c
// lte_dfts.c
void
init_fft
(
uint16_t
size
,
void
init_fft
(
uint16_t
size
,
uint8_t
logsize
,
uint8_t
logsize
,
...
...
openair1/PHY/defs.h
View file @
a0ffbdd7
...
@@ -580,6 +580,7 @@ typedef struct PHY_VARS_eNB_s {
...
@@ -580,6 +580,7 @@ typedef struct PHY_VARS_eNB_s {
time_stats_t
dlsch_rate_matching_stats
;
time_stats_t
dlsch_rate_matching_stats
;
time_stats_t
dlsch_turbo_encoding_stats
;
time_stats_t
dlsch_turbo_encoding_stats
;
time_stats_t
dlsch_interleaving_stats
;
time_stats_t
dlsch_interleaving_stats
;
time_stats_t
dl_ch_calib_stats
;
time_stats_t
ofdm_demod_stats
;
time_stats_t
ofdm_demod_stats
;
time_stats_t
rx_dft_stats
;
time_stats_t
rx_dft_stats
;
...
...
openair1/SIMULATION/LTE_PHY/dlsim.c
View file @
a0ffbdd7
...
@@ -1818,7 +1818,7 @@ int main(int argc, char **argv)
...
@@ -1818,7 +1818,7 @@ int main(int argc, char **argv)
if
((
transmission_mode
==
1
)
||
(
transmission_mode
==
7
))
{
if
((
transmission_mode
==
1
)
||
(
transmission_mode
==
7
))
{
for
(
aa
=
0
;
aa
<
eNB
->
frame_parms
.
nb_antennas_tx
;
aa
++
)
for
(
aa
=
0
;
aa
<
eNB
->
frame_parms
.
nb_antennas_tx
;
aa
++
)
for
(
re
=
0
;
re
<
eNB
->
frame_parms
.
ofdm_symbol_size
;
re
++
)
for
(
re
=
0
;
re
<
eNB
->
frame_parms
.
ofdm_symbol_size
;
re
++
)
eNB
->
common_vars
.
beam_weights
[
0
][
0
][
aa
][
re
]
=
0x00007fff
/
eNB
->
frame_parms
.
nb_antennas_tx
;
eNB
->
common_vars
.
beam_weights
[
0
][
0
][
aa
][
re
]
=
0x00007fff
/
sqrt
(
eNB
->
frame_parms
.
nb_antennas_tx
)
;
}
}
eNB
->
mac_enabled
=
1
;
eNB
->
mac_enabled
=
1
;
...
@@ -2320,7 +2320,7 @@ int main(int argc, char **argv)
...
@@ -2320,7 +2320,7 @@ int main(int argc, char **argv)
//PMI_FEEDBACK:
//PMI_FEEDBACK:
// printf("Trial %d : Round %d, pmi_feedback %d \n",trials,round,pmi_feedback);
// printf("Trial %d : Round %d, pmi_feedback %d \n",trials,round,pmi_feedback);
for
(
aa
=
0
;
aa
<
eNB
->
frame_parms
.
nb_antenna
s_tx
;
aa
++
)
{
for
(
aa
=
0
;
aa
<
eNB
->
frame_parms
.
nb_antenna
_ports_eNB
;
aa
++
)
{
memset
(
&
eNB
->
common_vars
.
txdataF
[
eNB_id
][
aa
][
0
],
0
,
FRAME_LENGTH_COMPLEX_SAMPLES_NO_PREFIX
*
sizeof
(
int32_t
));
memset
(
&
eNB
->
common_vars
.
txdataF
[
eNB_id
][
aa
][
0
],
0
,
FRAME_LENGTH_COMPLEX_SAMPLES_NO_PREFIX
*
sizeof
(
int32_t
));
}
}
...
...
openair1/SIMULATION/LTE_PHY/dlsim_tm7.c
View file @
a0ffbdd7
...
@@ -992,10 +992,6 @@ int main(int argc, char **argv)
...
@@ -992,10 +992,6 @@ int main(int argc, char **argv)
eNB
->
UE_stats
[
1
].
DL_pmi_single
=
0
;
eNB
->
UE_stats
[
1
].
DL_pmi_single
=
0
;
}
}
printf
(
"tdd_calib = %d
\n
"
,
tdd_calib
);
if
(
tdd_calib
==
1
)
//for first tests initialze calibration matrix with idendity
read_calibration_matrix
(
eNB
->
common_vars
.
tdd_calib_coeffs
[
0
],
"calibF.m"
,
frame_parms
);
if
(
input_fd
==
NULL
)
{
if
(
input_fd
==
NULL
)
{
...
@@ -2108,6 +2104,8 @@ int main(int argc, char **argv)
...
@@ -2108,6 +2104,8 @@ int main(int argc, char **argv)
initialize
(
&
time_vector_tx_mod
);
initialize
(
&
time_vector_tx_mod
);
struct
list
time_vector_tx_enc
;
struct
list
time_vector_tx_enc
;
initialize
(
&
time_vector_tx_enc
);
initialize
(
&
time_vector_tx_enc
);
struct
list
time_vector_tx_calib
;
initialize
(
&
time_vector_tx_calib
);
struct
list
time_vector_rx
;
struct
list
time_vector_rx
;
initialize
(
&
time_vector_rx
);
initialize
(
&
time_vector_rx
);
...
@@ -2148,6 +2146,7 @@ int main(int argc, char **argv)
...
@@ -2148,6 +2146,7 @@ int main(int argc, char **argv)
PMI_FEEDBACK:
PMI_FEEDBACK:
if
(
tdd_calib
==
1
)
{
if
(
tdd_calib
==
1
)
{
start_meas
(
&
eNB
->
dl_ch_calib_stats
);
//make sure dlsim is called with perfect channel estimation option (for freq_channel)
//make sure dlsim is called with perfect channel estimation option (for freq_channel)
//fill drs_ch_estimates with data from eNB2UE->chF
//fill drs_ch_estimates with data from eNB2UE->chF
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antenna_ports_eNB
;
aa
++
)
{
for
(
aa
=
0
;
aa
<
frame_parms
->
nb_antenna_ports_eNB
;
aa
++
)
{
...
@@ -2168,10 +2167,12 @@ PMI_FEEDBACK:
...
@@ -2168,10 +2167,12 @@ PMI_FEEDBACK:
eNB
->
common_vars
.
tdd_calib_coeffs
[
0
],
eNB
->
common_vars
.
tdd_calib_coeffs
[
0
],
frame_parms
);
frame_parms
);
/*
compute_BF_weights(eNB->dlsch[0][0]->ue_spec_bf_weights[0],
compute_BF_weights
(
eNB
->
dlsch
[
0
][
0
]
->
ue_spec_bf_weights
[
0
],
eNB
->
dlsch
[
0
][
0
]
->
calib_dl_ch_estimates
,
eNB
->
dlsch
[
0
][
0
]
->
calib_dl_ch_estimates
,
MRT
,
MRT
,
frame_parms);*/
frame_parms
);
stop_meas
(
&
eNB
->
dl_ch_calib_stats
);
}
}
//printf("Trial %d : Round %d, pmi_feedback %d \n",trials,round,pmi_feedback);
//printf("Trial %d : Round %d, pmi_feedback %d \n",trials,round,pmi_feedback);
...
@@ -3795,6 +3796,7 @@ PMI_FEEDBACK:
...
@@ -3795,6 +3796,7 @@ PMI_FEEDBACK:
double
t_tx_ifft
=
(
double
)
eNB
->
ofdm_mod_stats
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_tx_ifft
=
(
double
)
eNB
->
ofdm_mod_stats
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_tx_mod
=
(
double
)
eNB
->
dlsch_modulation_stats
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_tx_mod
=
(
double
)
eNB
->
dlsch_modulation_stats
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_tx_enc
=
(
double
)
eNB
->
dlsch_encoding_stats
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_tx_enc
=
(
double
)
eNB
->
dlsch_encoding_stats
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_tx_calib
=
(
double
)
eNB
->
dl_ch_calib_stats
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_rx
=
(
double
)
UE
->
phy_proc_rx
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
double
t_rx
=
(
double
)
UE
->
phy_proc_rx
.
p_time
/
cpu_freq_GHz
/
1000
.
0
;
...
@@ -3824,6 +3826,7 @@ PMI_FEEDBACK:
...
@@ -3824,6 +3826,7 @@ PMI_FEEDBACK:
push_front
(
&
time_vector_tx_ifft
,
t_tx_ifft
);
push_front
(
&
time_vector_tx_ifft
,
t_tx_ifft
);
push_front
(
&
time_vector_tx_mod
,
t_tx_mod
);
push_front
(
&
time_vector_tx_mod
,
t_tx_mod
);
push_front
(
&
time_vector_tx_enc
,
t_tx_enc
);
push_front
(
&
time_vector_tx_enc
,
t_tx_enc
);
push_front
(
&
time_vector_tx_calib
,
t_tx_calib
);
push_front
(
&
time_vector_rx
,
t_rx
);
push_front
(
&
time_vector_rx
,
t_rx
);
push_front
(
&
time_vector_rx_fft
,
t_rx_fft
);
push_front
(
&
time_vector_rx_fft
,
t_rx_fft
);
...
@@ -3842,6 +3845,8 @@ PMI_FEEDBACK:
...
@@ -3842,6 +3845,8 @@ PMI_FEEDBACK:
totable
(
table_tx_mod
,
&
time_vector_tx_mod
);
totable
(
table_tx_mod
,
&
time_vector_tx_mod
);
double
table_tx_enc
[
time_vector_tx_enc
.
size
];
double
table_tx_enc
[
time_vector_tx_enc
.
size
];
totable
(
table_tx_enc
,
&
time_vector_tx_enc
);
totable
(
table_tx_enc
,
&
time_vector_tx_enc
);
double
table_tx_calib
[
time_vector_tx_calib
.
size
];
totable
(
table_tx_calib
,
&
time_vector_tx_calib
);
double
table_rx
[
time_vector_rx
.
size
];
double
table_rx
[
time_vector_rx
.
size
];
totable
(
table_rx
,
&
time_vector_rx
);
totable
(
table_rx
,
&
time_vector_rx
);
...
@@ -3894,6 +3899,10 @@ PMI_FEEDBACK:
...
@@ -3894,6 +3899,10 @@ PMI_FEEDBACK:
double
tx_enc_q1
=
table_tx_enc
[
time_vector_tx_enc
.
size
/
4
];
double
tx_enc_q1
=
table_tx_enc
[
time_vector_tx_enc
.
size
/
4
];
double
tx_enc_q3
=
table_tx_enc
[
3
*
time_vector_tx_enc
.
size
/
4
];
double
tx_enc_q3
=
table_tx_enc
[
3
*
time_vector_tx_enc
.
size
/
4
];
double
tx_calib_median
=
table_tx_calib
[
time_vector_tx_calib
.
size
/
2
];
double
tx_calib_q1
=
table_tx_calib
[
time_vector_tx_calib
.
size
/
4
];
double
tx_calib_q3
=
table_tx_calib
[
3
*
time_vector_tx_calib
.
size
/
4
];
double
rx_median
=
table_rx
[
time_vector_rx
.
size
/
2
];
double
rx_median
=
table_rx
[
time_vector_rx
.
size
/
2
];
double
rx_q1
=
table_rx
[
time_vector_rx
.
size
/
4
];
double
rx_q1
=
table_rx
[
time_vector_rx
.
size
/
4
];
double
rx_q3
=
table_rx
[
3
*
time_vector_rx
.
size
/
4
];
double
rx_q3
=
table_rx
[
3
*
time_vector_rx
.
size
/
4
];
...
@@ -3914,6 +3923,7 @@ PMI_FEEDBACK:
...
@@ -3914,6 +3923,7 @@ PMI_FEEDBACK:
double
std_phy_proc_tx_ifft
=
0
;
double
std_phy_proc_tx_ifft
=
0
;
double
std_phy_proc_tx_mod
=
0
;
double
std_phy_proc_tx_mod
=
0
;
double
std_phy_proc_tx_enc
=
0
;
double
std_phy_proc_tx_enc
=
0
;
double
std_phy_proc_tx_calib
=
0
;
double
std_phy_proc_rx
=
0
;
double
std_phy_proc_rx
=
0
;
double
std_phy_proc_rx_fft
=
0
;
double
std_phy_proc_rx_fft
=
0
;
...
@@ -3984,6 +3994,10 @@ PMI_FEEDBACK:
...
@@ -3984,6 +3994,10 @@ PMI_FEEDBACK:
printf
(
"|__ DLSCH sub-block interleaving time :%f us (%d trials)
\n
"
,
printf
(
"|__ DLSCH sub-block interleaving time :%f us (%d trials)
\n
"
,
((
double
)
eNB
->
dlsch_interleaving_stats
.
trials
/
eNB
->
dlsch_encoding_stats
.
trials
)
*
(
double
)
((
double
)
eNB
->
dlsch_interleaving_stats
.
trials
/
eNB
->
dlsch_encoding_stats
.
trials
)
*
(
double
)
eNB
->
dlsch_interleaving_stats
.
diff
/
eNB
->
dlsch_interleaving_stats
.
trials
/
cpu_freq_GHz
/
1000
.
0
,
eNB
->
dlsch_interleaving_stats
.
trials
);
eNB
->
dlsch_interleaving_stats
.
diff
/
eNB
->
dlsch_interleaving_stats
.
trials
/
cpu_freq_GHz
/
1000
.
0
,
eNB
->
dlsch_interleaving_stats
.
trials
);
std_phy_proc_tx_calib
=
sqrt
((
double
)
eNB
->
dl_ch_calib_stats
.
diff_square
/
pow
(
cpu_freq_GHz
,
2
)
/
pow
(
1000
,
2
)
/
eNB
->
dl_ch_calib_stats
.
trials
-
pow
((
double
)
eNB
->
dl_ch_calib_stats
.
diff
/
eNB
->
dl_ch_calib_stats
.
trials
/
cpu_freq_GHz
/
1000
,
2
));
printf
(
"TDD_DL_calib time :%f us (%d trials)
\n
"
,(
double
)
eNB
->
dl_ch_calib_stats
.
diff
/
eNB
->
dl_ch_calib_stats
.
trials
/
cpu_freq_GHz
/
1000
.
0
,
eNB
->
dl_ch_calib_stats
.
trials
);
printf
(
"|__ Statistcs std: %fus median %fus q1 %fus q3 %fus
\n
"
,
std_phy_proc_tx_calib
,
tx_calib_median
,
tx_calib_q1
,
tx_calib_q3
);
printf
(
"
\n\n
UE RX function statistics (per 1ms subframe)
\n\n
"
);
printf
(
"
\n\n
UE RX function statistics (per 1ms subframe)
\n\n
"
);
std_phy_proc_rx
=
sqrt
((
double
)
UE
->
phy_proc_rx
.
diff_square
/
pow
(
cpu_freq_GHz
,
2
)
/
pow
(
1000
,
std_phy_proc_rx
=
sqrt
((
double
)
UE
->
phy_proc_rx
.
diff_square
/
pow
(
cpu_freq_GHz
,
2
)
/
pow
(
1000
,
...
...
targets/RT/USER/lte-softmodem.c
View file @
a0ffbdd7
...
@@ -1627,7 +1627,8 @@ int main( int argc, char **argv )
...
@@ -1627,7 +1627,8 @@ int main( int argc, char **argv )
if
((
transmission_mode
==
1
)
||
(
transmission_mode
==
7
))
{
if
((
transmission_mode
==
1
)
||
(
transmission_mode
==
7
))
{
for
(
j
=
0
;
j
<
frame_parms
[
CC_id
]
->
nb_antennas_tx
;
j
++
)
for
(
j
=
0
;
j
<
frame_parms
[
CC_id
]
->
nb_antennas_tx
;
j
++
)
for
(
re
=
0
;
re
<
frame_parms
[
CC_id
]
->
ofdm_symbol_size
;
re
++
)
for
(
re
=
0
;
re
<
frame_parms
[
CC_id
]
->
ofdm_symbol_size
;
re
++
)
PHY_vars_eNB_g
[
0
][
CC_id
]
->
common_vars
.
beam_weights
[
0
][
0
][
j
][
re
]
=
0x00007fff
/
frame_parms
[
CC_id
]
->
nb_antennas_tx
;
//In softmodem: the power constraint is on each antenna, so we do not norm the beam weights
PHY_vars_eNB_g
[
0
][
CC_id
]
->
common_vars
.
beam_weights
[
0
][
0
][
j
][
re
]
=
0x00007fff
;
///sqrt(frame_parms[CC_id]->nb_antennas_tx);
}
}
if
(
phy_test
==
1
)
PHY_vars_eNB_g
[
0
][
CC_id
]
->
mac_enabled
=
0
;
if
(
phy_test
==
1
)
PHY_vars_eNB_g
[
0
][
CC_id
]
->
mac_enabled
=
0
;
else
PHY_vars_eNB_g
[
0
][
CC_id
]
->
mac_enabled
=
1
;
else
PHY_vars_eNB_g
[
0
][
CC_id
]
->
mac_enabled
=
1
;
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment