Fix: SNR feedback to PUSCH TPC and PUCCH TPC of default scheduler.

This commit is contained in:
Haruki NAOI
2020-11-30 10:55:19 +09:00
committed by shono.takafumi
parent de568b2d88
commit 1a320bc6d4
2 changed files with 2 additions and 3 deletions
@@ -239,7 +239,6 @@ void schedule_ulsch_phy_test(module_id_t module_idP,frame_t frameP,sub_frame_t s
//compute the expected ULSCH RX power (for the stats)
// this is the snr and this should be constant (regardless of mcs)
snr = UE_sched_ctrl->pusch_snr[CC_id];
// new transmission
ndi = 1-UE_template->oldNDI_UL[harq_pid];
UE_template->oldNDI_UL[harq_pid]=ndi;
+2 -2
View File
@@ -1492,7 +1492,7 @@ schedule_ulsch_rnti(module_id_t module_idP,
* This is the normalized RX snr and this should be constant (regardless
* of mcs) Is not in dBm, unit from nfapi, converting to dBm
*/
const int32_t snr = (5 * UE_sched_ctrl_ptr->pusch_snr[CC_id] - 640) / 10;
const int32_t snr = UE_sched_ctrl_ptr->pusch_snr[CC_id];
const int32_t target_snr = mac->puSch10xSnr / 10;
/*
@@ -2073,7 +2073,7 @@ void schedule_ulsch_rnti_emtc(module_id_t module_idP,
cqi_req = 0;
/* Power control: compute the expected ULSCH RX snr (for the stats) */
/* This is the normalized snr and this should be constant (regardless of mcs) */
snr = UE_sched_ctrl->pusch_snr_avg[CC_id];
snr = UE_sched_ctrl->pusch_snr[CC_id];
target_snr = eNB->puSch10xSnr / 10; /* TODO: target_rx_power was 178, what to put? */
/* This assumes accumulated tpc */
/* Make sure that we are only sending a tpc update once a frame, otherwise the control loop will freak out */