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
zzha zzha
OpenXG-RAN
Commits
8bf7a13f
Commit
8bf7a13f
authored
Feb 26, 2020
by
Robert Schmidt
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
Remove first_rb from ULSCH scheduler
parent
8a84e2d9
Changes
4
Show whitespace changes
Inline
Side-by-side
Showing
4 changed files
with
72 additions
and
101 deletions
+72
-101
openair2/LAYER2/MAC/eNB_scheduler_ulsch.c
openair2/LAYER2/MAC/eNB_scheduler_ulsch.c
+47
-84
openair2/LAYER2/MAC/mac.h
openair2/LAYER2/MAC/mac.h
+2
-1
openair2/LAYER2/MAC/mac_proto.h
openair2/LAYER2/MAC/mac_proto.h
+3
-5
openair2/LAYER2/MAC/pre_processor.c
openair2/LAYER2/MAC/pre_processor.c
+20
-11
No files found.
openair2/LAYER2/MAC/eNB_scheduler_ulsch.c
View file @
8bf7a13f
...
...
@@ -1099,14 +1099,12 @@ schedule_ulsch(module_id_t module_idP,
sub_frame_t
subframeP
)
//-----------------------------------------------------------------------------
{
uint16_t
first_rb
[
NFAPI_CC_MAX
];
eNB_MAC_INST
*
mac
=
NULL
;
COMMON_channels_t
*
cc
=
NULL
;
int
sched_subframe
;
int
sched_frame
;
/* Init */
mac
=
RC
.
mac
[
module_idP
];
memset
(
first_rb
,
0
,
NFAPI_CC_MAX
*
sizeof
(
uint16_t
));
start_meas
(
&
(
mac
->
schedule_ulsch
));
sched_subframe
=
(
subframeP
+
4
)
%
10
;
sched_frame
=
frameP
;
...
...
@@ -1213,9 +1211,6 @@ schedule_ulsch(module_id_t module_idP,
/* Note: RC.nb_mac_CC[module_idP] should be lower than or equal to NFAPI_CC_MAX */
for
(
int
CC_id
=
0
;
CC_id
<
RC
.
nb_mac_CC
[
module_idP
];
CC_id
++
,
cc
++
)
{
first_rb
[
CC_id
]
=
(
emtc_active
[
CC_id
]
==
1
)
?
7
:
1
;
RA_t
*
ra_ptr
=
cc
->
ra
;
/* From Louis-Adrien to François:
* The comment bloc below is to configure with a command line.
* I took it from the equivalent part in the fairRR scheduler (around line 2578 in eNB_scheduler_fairRR.c).
...
...
@@ -1226,6 +1221,7 @@ schedule_ulsch(module_id_t module_idP,
* I think it should be sched_frame instead. This parameter has only impacts in case TDD and preamble format 4.
* To confirm.
*/
/* TODO: update vrb_map_UL here? */
/*
int start_rb = 0;
int nb_rb = 6;
...
...
@@ -1242,27 +1238,41 @@ schedule_ulsch(module_id_t module_idP,
}
*/
/*
* Check if RA (Msg3) is active in this subframeP, if so skip the PRB used for Msg3
* Msg3 is using 1 PRB so we need to increase first_rb accordingly
* Not sure about the break (can there be more than 1 active RA procedure per CC_id and per subframe?)
/* HACK: let's remove the PUCCH from available RBs
* we suppose PUCCH size is:
* - for 25 RBs: 1 RB (top and bottom of ressource grid)
* - for 50: 2 RBs
* - for 100: 3 RBs
* This is totally arbitrary and might even be wrong.
*/
for
(
int
ra_index
=
0
;
ra_index
<
NB_RA_PROC_MAX
;
ra_index
++
,
ra_ptr
++
)
{
if
((
ra_ptr
->
state
==
WAITMSG3
)
&&
(
ra_ptr
->
Msg3_subframe
==
sched_subframe
))
{
if
(
first_rb
[
CC_id
]
<
ra_ptr
->
msg3_first_rb
+
ra_ptr
->
msg3_nb_rb
)
{
first_rb
[
CC_id
]
=
ra_ptr
->
msg3_first_rb
+
ra_ptr
->
msg3_nb_rb
;
}
switch
(
to_prb
(
cc
[
CC_id
].
ul_Bandwidth
)
)
{
case
25
:
cc
[
CC_id
].
vrb_map_UL
[
0
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
24
]
=
1
;
break
;
/* Louis-Adrien: I couldn't find an interdiction of multiple Msg3 scheduling
* on the same time resources. Also the performance improvement of breaking is low,
* since we will loop until the end, most of the time.
* I'm letting the break as a reminder, in case of misunderstanding the spec.
*/
// break;
}
case
50
:
cc
[
CC_id
].
vrb_map_UL
[
0
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
1
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
48
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
49
]
=
1
;
break
;
case
100
:
cc
[
CC_id
].
vrb_map_UL
[
0
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
1
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
2
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
97
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
98
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
99
]
=
1
;
break
;
default:
LOG_E
(
MAC
,
"RBs setting not handled. Todo.
\n
"
);
exit
(
1
);
}
schedule_ulsch_rnti
(
module_idP
,
CC_id
,
frameP
,
subframeP
,
sched_subframe
,
first_rb
);
schedule_ulsch_rnti
(
module_idP
,
CC_id
,
frameP
,
subframeP
,
sched_subframe
);
}
stop_meas
(
&
mac
->
schedule_ulsch
);
...
...
@@ -1277,8 +1287,7 @@ schedule_ulsch_rnti(module_id_t module_idP,
int
CC_id
,
frame_t
frameP
,
sub_frame_t
subframeP
,
unsigned
char
sched_subframeP
,
uint16_t
*
first_rb
)
{
unsigned
char
sched_subframeP
)
{
const
uint8_t
aggregation
=
2
;
/* TODO: does this need to be static? */
static
int32_t
tpc_accumulated
=
0
;
...
...
@@ -1306,43 +1315,13 @@ schedule_ulsch_rnti(module_id_t module_idP,
nfapi_ul_config_ulsch_harq_information
*
ulsch_harq_information
;
hi_dci0_req
->
sfn_sf
=
(
frameP
<<
4
)
+
subframeP
;
int
n_rb_ul_tab
=
to_prb
(
cc
[
CC_id
].
ul_Bandwidth
);
/* HACK: let's remove the PUCCH from available RBs
* we suppose PUCCH size is:
* - for 25 RBs: 1 RB (top and bottom of ressource grid)
* - for 50: 2 RBs
* - for 100: 3 RBs
* This is totally arbitrary and might even be wrong.
* We suppose 'first_rb[]' has been correctly populated by the caller,
* so we only remove the top part of the resource grid.
*/
switch
(
n_rb_ul_tab
)
{
case
25
:
n_rb_ul_tab
-=
1
;
break
;
case
50
:
n_rb_ul_tab
-=
2
;
break
;
case
100
:
n_rb_ul_tab
-=
3
;
break
;
default:
LOG_E
(
MAC
,
"RBs setting not handled. Todo.
\n
"
);
exit
(
1
);
}
UE_info
->
first_rb_offset
[
CC_id
]
=
n_rb_ul_tab
;
/*
* ULSCH preprocessor: set UE_template->
* pre_allocated_nb_rb_ul[slice_idx]
* pre_assigned_mcs_ul
* pre_allocated_rb_table_index_ul
*/
ulsch_scheduler_pre_processor
(
module_idP
,
CC_id
,
frameP
,
subframeP
,
sched_frame
,
sched_subframeP
,
first_rb
);
ulsch_scheduler_pre_processor
(
module_idP
,
CC_id
,
frameP
,
subframeP
,
sched_frame
,
sched_subframeP
);
for
(
int
UE_id
=
UE_info
->
list
.
head
;
UE_id
>=
0
;
UE_id
=
UE_info
->
list
.
next
[
UE_id
])
{
if
(
UE_info
->
UE_template
[
CC_id
][
UE_id
].
rach_resource_type
>
0
)
...
...
@@ -1382,9 +1361,8 @@ schedule_ulsch_rnti(module_id_t module_idP,
UE_id
,
rnti
);
LOG_D
(
MAC
,
"[eNB %d] %d.%d (sched_frame %d, sched_subframe %d), "
"Checking PUSCH %d for UE %d/%x CC %d : aggregation level %d, "
"N_RB_UL %d
\n
"
,
"[eNB %d] %d.%d (sched %d.%d), "
"Checking PUSCH %d for UE %d/%x CC %d : aggregation level %d
\n
"
,
module_idP
,
frameP
,
subframeP
,
...
...
@@ -1394,8 +1372,7 @@ schedule_ulsch_rnti(module_id_t module_idP,
UE_id
,
rnti
,
CC_id
,
aggregation
,
n_rb_ul_tab
);
aggregation
);
/* Seems unused, only for debug */
RC
.
eNB
[
module_idP
][
CC_id
]
->
pusch_stats_BO
[
UE_id
][(
frameP
*
10
)
+
subframeP
]
=
UE_template_ptr
->
estimated_ul_buffer
;
...
...
@@ -1570,11 +1547,8 @@ schedule_ulsch_rnti(module_id_t module_idP,
UE_info
->
eNB_UE_stats
[
CC_id
][
UE_id
].
ulsch_mcs2
=
UE_template_ptr
->
mcs_UL
[
harq_pid
];
while
(((
rb_table
[
rb_table_index
]
>
(
n_rb_ul_tab
-
first_rb
[
CC_id
]))
||
(
rb_table
[
rb_table_index
]
>
45
))
&&
(
rb_table_index
>
0
))
{
while
(
rb_table
[
rb_table_index
]
>
45
&&
rb_table_index
>
0
)
rb_table_index
--
;
}
UE_template_ptr
->
TBS_UL
[
harq_pid
]
=
get_TBS_UL
(
UE_template_ptr
->
mcs_UL
[
harq_pid
],
rb_table
[
rb_table_index
]);
UE_info
->
eNB_UE_stats
[
CC_id
][
UE_id
].
total_rbs_used_rx
+=
rb_table
[
rb_table_index
];
...
...
@@ -1588,13 +1562,12 @@ schedule_ulsch_rnti(module_id_t module_idP,
T_INT
(
subframeP
),
T_INT
(
harq_pid
),
T_INT
(
UE_template_ptr
->
mcs_UL
[
harq_pid
]),
T_INT
(
first_rb
[
CC_id
]),
T_INT
(
rb_table
[
rb_table_index
]),
T_INT
(
UE_template_ptr
->
TBS_UL
[
harq_pid
]),
T_INT
(
ndi
));
/* Store information for possible retransmission */
UE_template_ptr
->
nb_rb_ul
[
harq_pid
]
=
rb_table
[
rb_table_index
];
UE_template_ptr
->
first_rb_ul
[
harq_pid
]
=
first_rb
[
CC_id
]
;
UE_template_ptr
->
first_rb_ul
[
harq_pid
]
=
UE_template_ptr
->
pre_first_nb_rb_ul
;
UE_template_ptr
->
cqi_req
[
harq_pid
]
=
cqi_req
;
UE_sched_ctrl_ptr
->
ul_scheduled
|=
(
1
<<
harq_pid
);
...
...
@@ -1632,7 +1605,7 @@ schedule_ulsch_rnti(module_id_t module_idP,
hi_dci0_pdu
->
dci_pdu
.
dci_pdu_rel8
.
aggregation_level
=
aggregation
;
hi_dci0_pdu
->
dci_pdu
.
dci_pdu_rel8
.
rnti
=
rnti
;
hi_dci0_pdu
->
dci_pdu
.
dci_pdu_rel8
.
transmission_power
=
6000
;
hi_dci0_pdu
->
dci_pdu
.
dci_pdu_rel8
.
resource_block_start
=
first_rb
[
CC_id
]
;
hi_dci0_pdu
->
dci_pdu
.
dci_pdu_rel8
.
resource_block_start
=
UE_template_ptr
->
pre_first_nb_rb_ul
;
hi_dci0_pdu
->
dci_pdu
.
dci_pdu_rel8
.
number_of_resource_block
=
rb_table
[
rb_table_index
];
hi_dci0_pdu
->
dci_pdu
.
dci_pdu_rel8
.
mcs_1
=
...
...
@@ -1688,7 +1661,7 @@ schedule_ulsch_rnti(module_id_t module_idP,
get_tmode
(
module_idP
,
CC_id
,
UE_id
),
mac
->
ul_handle
,
rnti
,
first_rb
[
CC_id
]
,
// resource_block_start
UE_template_ptr
->
pre_first_nb_rb_ul
,
// resource_block_start
rb_table
[
rb_table_index
],
// number_of_resource_blocks
UE_template_ptr
->
mcs_UL
[
harq_pid
],
cshift
,
// cyclic_shift_2_for_drms
...
...
@@ -1757,8 +1730,6 @@ schedule_ulsch_rnti(module_id_t module_idP,
cqi_req
,
UE_id
,
rnti
);
/* Increment first rb for next UE allocation */
first_rb
[
CC_id
]
+=
rb_table
[
rb_table_index
];
}
else
{
// round_index > 0 => retransmission
T
(
T_ENB_MAC_UE_UL_SCHEDULE_RETRANSMISSION
,
T_INT
(
module_idP
),
...
...
@@ -1868,20 +1839,6 @@ schedule_ulsch_rnti(module_id_t module_idP,
sched_frame
,
sched_subframeP
,
cqi_req
);
/* HACK: RBs used by retransmission have to be reserved.
* The current mechanism uses the notion of 'first_rb', so
* we skip all RBs below the ones retransmitted. This is
* not correct. Imagine only RB 23 is retransmitted, then all
* RBs < 23 will be marked unusable for new transmissions (case where
* round == 0). Note also that this code works only if the preprocessor
* orders UEs with retransmission with higher priority than UEs with new
* transmission.
* All this should be cleaned up properly.
*/
if
(
first_rb
[
CC_id
]
<
UE_template_ptr
->
first_rb_ul
[
harq_pid
]
+
UE_template_ptr
->
nb_rb_ul
[
harq_pid
])
first_rb
[
CC_id
]
=
UE_template_ptr
->
first_rb_ul
[
harq_pid
]
+
UE_template_ptr
->
nb_rb_ul
[
harq_pid
];
}
// end of round > 0
}
// loop over UE_ids
}
...
...
@@ -2018,6 +1975,12 @@ void schedule_ulsch_rnti_emtc(module_id_t module_idP,
UE_sched_ctrl
->
cqi_req_timer
);
/* Reset the scheduling request */
emtc_active
[
CC_id
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
1
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
2
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
3
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
4
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
5
]
=
1
;
cc
[
CC_id
].
vrb_map_UL
[
6
]
=
1
;
UE_template
->
ul_SR
=
0
;
status
=
mac_eNB_get_rrc_status
(
module_idP
,
rnti
);
cqi_req
=
0
;
...
...
openair2/LAYER2/MAC/mac.h
View file @
8bf7a13f
...
...
@@ -825,6 +825,8 @@ typedef struct {
/// Number of Allocated RBs by the ulsch preprocessor
uint8_t
pre_allocated_nb_rb_ul
;
/// Start of Allocated RBs by the USLCH preprocessor
uint8_t
pre_first_nb_rb_ul
;
/// index of Allocated RBs by the ulsch preprocessor
int8_t
pre_allocated_rb_table_index_ul
;
...
...
@@ -1147,7 +1149,6 @@ typedef struct {
/// Sorting criteria for the UE list in the MAC preprocessor
uint16_t
sorting_criteria
[
MAX_NUM_SLICES
][
CR_NUM
];
uint16_t
first_rb_offset
[
NFAPI_CC_MAX
];
}
UE_info_t
;
/*! \brief deleting control information*/
...
...
openair2/LAYER2/MAC/mac_proto.h
View file @
8bf7a13f
...
...
@@ -129,8 +129,7 @@ void schedule_ulsch(module_id_t module_idP, frame_t frameP,
*/
void
schedule_ulsch_rnti
(
module_id_t
module_idP
,
int
CC_id
,
frame_t
frameP
,
sub_frame_t
subframe
,
unsigned
char
sched_subframe
,
uint16_t
*
first_rb
);
unsigned
char
sched_subframe
);
void
schedule_ulsch_rnti_emtc
(
module_id_t
module_idP
,
frame_t
frameP
,
...
...
@@ -674,15 +673,14 @@ void ulsch_scheduler_pre_processor(module_id_t module_idP,
int
frameP
,
sub_frame_t
subframeP
,
int
sched_frameP
,
unsigned
char
sched_subframeP
,
uint16_t
*
first_rb
);
unsigned
char
sched_subframeP
);
void
store_ulsch_buffer
(
module_id_t
module_idP
,
int
frameP
,
sub_frame_t
subframeP
);
void
assign_max_mcs_min_rb
(
module_id_t
module_idP
,
int
CC_id
,
int
frameP
,
sub_frame_t
subframeP
,
uint16_t
*
first_rb
);
int
available_rbs
);
void
adjust_bsr_info
(
int
buffer_occupancy
,
uint16_t
TBS
,
UE_TEMPLATE
*
UE_template
);
...
...
openair2/LAYER2/MAC/pre_processor.c
View file @
8bf7a13f
...
...
@@ -391,21 +391,30 @@ void ulsch_scheduler_pre_processor(module_id_t module_idP,
int
frameP
,
sub_frame_t
subframeP
,
int
sched_frameP
,
unsigned
char
sched_subframeP
,
uint16_t
*
first_rb
)
{
unsigned
char
sched_subframeP
)
{
uint16_t
nb_allocated_rbs
[
MAX_MOBILES_PER_ENB
];
uint16_t
total_allocated_rbs
=
0
;
uint16_t
average_rbs_per_user
=
0
;
int16_t
total_remaining_rbs
=
0
;
uint16_t
total_ue_count
=
0
;
eNB_MAC_INST
*
eNB
=
RC
.
mac
[
module_idP
];
UE_info_t
*
UE_info
=
&
eNB
->
UE_info
;
const
int
N_RB_UL
=
to_prb
(
eNB
->
common_channels
[
CC_id
].
ul_Bandwidth
);
uint16_t
available_rbs
=
N_RB_UL
-
2
*
first_rb
[
CC_id
];
// top and bottom // - UE_info->first_rb_offset[CC_id];
// maximize MCS and then allocate required RB according to the buffer occupancy with the limit of max available UL RB
UE_info_t
*
UE_info
=
&
RC
.
mac
[
module_idP
]
->
UE_info
;
const
int
N_RB_UL
=
to_prb
(
RC
.
mac
[
module_idP
]
->
common_channels
[
CC_id
].
ul_Bandwidth
);
const
COMMON_channels_t
*
cc
=
&
RC
.
mac
[
module_idP
]
->
common_channels
[
CC_id
];
int
available_rbs
=
0
;
int
first_rb
=
-
1
;
for
(
int
i
=
0
;
i
<
N_RB_UL
;
++
i
)
{
if
(
cc
->
vrb_map_UL
[
i
]
==
0
)
{
available_rbs
++
;
if
(
first_rb
<
0
)
first_rb
=
i
;
}
}
// maximize MCS and then allocate required RB according to the buffer
// occupancy with the limit of max available UL RB
LOG_D
(
MAC
,
"In ulsch_preprocessor: assign max mcs min rb
\n
"
);
assign_max_mcs_min_rb
(
module_idP
,
CC_id
,
frameP
,
subframeP
,
first_rb
);
assign_max_mcs_min_rb
(
module_idP
,
CC_id
,
frameP
,
subframeP
,
available_rbs
);
for
(
int
UE_id
=
UE_info
->
list
.
head
;
UE_id
>=
0
;
UE_id
=
UE_info
->
list
.
next
[
UE_id
])
{
if
(
UE_info
->
UE_template
[
CC_id
][
UE_id
].
pre_allocated_nb_rb_ul
>
0
)
{
...
...
@@ -462,7 +471,9 @@ void ulsch_scheduler_pre_processor(module_id_t module_idP,
total_allocated_rbs
++
;
}
UE_template
->
pre_first_nb_rb_ul
=
first_rb
;
UE_template
->
pre_allocated_nb_rb_ul
=
nb_allocated_rbs
[
UE_id
];
first_rb
+=
nb_allocated_rbs
[
UE_id
];
LOG_D
(
MAC
,
"******************UL Scheduling Information for UE%d CC_id %d ************************
\n
"
,
UE_id
,
CC_id
);
...
...
@@ -524,9 +535,7 @@ assign_max_mcs_min_rb(module_id_t module_idP,
int
CC_id
,
int
frameP
,
sub_frame_t
subframeP
,
uint16_t
*
first_rb
)
{
const
int
N_RB_UL
=
to_prb
(
RC
.
mac
[
module_idP
]
->
common_channels
[
CC_id
].
ul_Bandwidth
);
const
int
available_rbs
=
N_RB_UL
-
2
*
first_rb
[
CC_id
];
// top and bottom - UE_info->first_rb_offset[CC_id];
int
available_rbs
)
{
const
int
Ncp
=
RC
.
mac
[
module_idP
]
->
common_channels
[
CC_id
].
Ncp
;
UE_info_t
*
UE_info
=
&
RC
.
mac
[
module_idP
]
->
UE_info
;
...
...
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