Change in osmo-bts[master]: MS Power Control Loop: Fix oscillations within good MS Power Levels

This is merely a historical archive of years 2008-2021, before the migration to mailman3.

A maintained and still updated list archive can be found at https://lists.osmocom.org/hyperkitty/list/gerrit-log@lists.osmocom.org/.

pespin gerrit-no-reply at lists.osmocom.org
Fri Sep 17 12:23:56 UTC 2021


pespin has uploaded this change for review. ( https://gerrit.osmocom.org/c/osmo-bts/+/25503 )


Change subject: MS Power Control Loop: Fix oscillations within good MS Power Levels
......................................................................

MS Power Control Loop: Fix oscillations within good MS Power Levels

Related: SYS#4917
Change-Id: Ib33a94eba92004f9327c91e020a506dfa9a957c3
---
M src/common/power_control.c
M tests/power/ms_power_loop_test.c
M tests/power/ms_power_loop_test.err
M tests/power/ms_power_loop_test.ok
4 files changed, 44 insertions(+), 19 deletions(-)



  git pull ssh://gerrit.osmocom.org:29418/osmo-bts refs/changes/03/25503/1

diff --git a/src/common/power_control.c b/src/common/power_control.c
index 62905bd..a29d6db 100644
--- a/src/common/power_control.c
+++ b/src/common/power_control.c
@@ -199,6 +199,7 @@
 	uint8_t rxlev_avg;
 	int16_t ul_lqual_cb_avg;
 	const struct gsm_power_ctrl_meas_params *ci_meas;
+	bool ignore;
 
 	if (!trx_ms_pwr_ctrl_is_osmo(trx))
 		return 0;
@@ -265,15 +266,39 @@
 	}
 
 	current_dbm = ms_pwr_dbm(band, state->current);
+
+	/* In this Power Control Loop, we infer a new good MS Power Level based
+	 * on the previous MS Power Level announced by the MS (not the previous
+	 * one we requested!) together with the related computed measurements.
+	 * Hence, and since we allow for several good MS Power Levels falling into our
+	 * thresholds, we could finally converge into an oscillation loop where
+	 * the MS bounces between 2 different correct MS Power levels all the
+	 * time, due to the fact that we "accept" and "request back" whatever
+	 * good MS Power Level we received from the MS, but at that time the MS
+	 * will be transmitting using the previous MS Powever Level we
+	 * requested, which we will later "accept" and "request back" on next loop
+	 * iteration. As a result MS effectively bounces between those 2 MS
+	 * Power Levels.
+	 * In order to fix this permanent oscillation, if current MS_PWR used/announced
+	 * by MS is good ("ms_dbm == new_dbm", hence within thresholds and no change
+	 * required) but has higher Tx power than the one we last requested, we ignore
+	 * it and keep requesting for one with lower Tx power. This way we converge to
+	 * the lowest good Tx power avoiding oscillating over values within thresholds.
+	 */
+	ignore = (ms_dbm == new_dbm && ms_dbm > current_dbm);
+
 	LOGPLCHAN(lchan, DLOOP, LOGL_INFO, "%s MS power control level %d (%d dBm) => %d (%d dBm): "
 		  "ms-pwr-lvl[curr %" PRIu8 ", max %" PRIu8 "], RSSI[curr %d, avg %d, thresh %d..%d] dBm,"
 		  " C/I[curr %d, avg %d, thresh %d..%d] dB\n",
-		  (new_dbm > current_dbm) ? "Raising" : "Lowering",
+		  ignore ? "Ignoring" : ((new_dbm > current_dbm) ? "Raising" : "Lowering"),
 		  state->current, current_dbm, new_power_lvl, new_dbm, ms_power_lvl,
 		  state->max, ul_rssi_dbm, rxlev2dbm(rxlev_avg),
 		  rxlev2dbm(params->rxlev_meas.lower_thresh), rxlev2dbm(params->rxlev_meas.upper_thresh),
 		  ul_lqual_cb/10, ul_lqual_cb_avg/10, ci_meas->lower_thresh, ci_meas->upper_thresh);
 
+	if (ignore)
+		return 0;
+
 	/* store the resulting new MS power level in the lchan */
 	state->current = new_power_lvl;
 	bts_model_adjst_ms_pwr(lchan);
diff --git a/tests/power/ms_power_loop_test.c b/tests/power/ms_power_loop_test.c
index d8d1c71..a343ee2 100644
--- a/tests/power/ms_power_loop_test.c
+++ b/tests/power/ms_power_loop_test.c
@@ -398,11 +398,11 @@
 	lchan->ms_power_ctrl.current = 10;
 	lchan->ms_power_ctrl.max = 2;
 
-	apply_power_test_ext(lchan, 9, good_rxlev, good_lqual, 1, 9);
-	apply_power_test_ext(lchan, 10, good_rxlev, good_lqual, 1, 10);
-	apply_power_test_ext(lchan, 9, good_rxlev, good_lqual, 1, 9);
-	apply_power_test_ext(lchan, 10, good_rxlev, good_lqual, 1, 10);
-	apply_power_test_ext(lchan, 9, good_rxlev, good_lqual, 1, 9);
+	apply_power_test_ext(lchan, 9, good_rxlev, good_lqual, 1, 10);
+	apply_power_test_ext(lchan, 10, good_rxlev, good_lqual, 0, 10);
+	apply_power_test_ext(lchan, 9, good_rxlev, good_lqual, 0, 10);
+	apply_power_test_ext(lchan, 10, good_rxlev, good_lqual, 0, 10);
+	apply_power_test_ext(lchan, 9, good_rxlev, good_lqual, 0, 10);
 }
 
 int main(int argc, char **argv)
diff --git a/tests/power/ms_power_loop_test.err b/tests/power/ms_power_loop_test.err
index 5a68e3b..a85b9b2 100644
--- a/tests/power/ms_power_loop_test.err
+++ b/tests/power/ms_power_loop_test.err
@@ -58,8 +58,8 @@
 (bts=0,trx=0,ts=0,ss=0) Lowering MS power control level 11 (8 dBm) => 12 (6 dBm): ms-pwr-lvl[curr 11, max 2], RSSI[curr -100, avg -100, thresh -75..-75] dBm, C/I[curr 17, avg 17, thresh 12..16] dB
 (bts=0,trx=0,ts=0,ss=0) Lowering MS power control level 12 (6 dBm) => 13 (4 dBm): ms-pwr-lvl[curr 12, max 2], RSSI[curr -60, avg -60, thresh -75..-75] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
 (bts=0,trx=0,ts=0,ss=0) Lowering MS power control level 13 (4 dBm) => 14 (2 dBm): ms-pwr-lvl[curr 13, max 2], RSSI[curr -60, avg -60, thresh -75..-75] dBm, C/I[curr 17, avg 17, thresh 12..16] dB
-(bts=0,trx=0,ts=0,ss=0) Raising MS power control level 10 (10 dBm) => 9 (12 dBm): ms-pwr-lvl[curr 9, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
-(bts=0,trx=0,ts=0,ss=0) Lowering MS power control level 9 (12 dBm) => 10 (10 dBm): ms-pwr-lvl[curr 10, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
-(bts=0,trx=0,ts=0,ss=0) Raising MS power control level 10 (10 dBm) => 9 (12 dBm): ms-pwr-lvl[curr 9, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
-(bts=0,trx=0,ts=0,ss=0) Lowering MS power control level 9 (12 dBm) => 10 (10 dBm): ms-pwr-lvl[curr 10, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
-(bts=0,trx=0,ts=0,ss=0) Raising MS power control level 10 (10 dBm) => 9 (12 dBm): ms-pwr-lvl[curr 9, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
+(bts=0,trx=0,ts=0,ss=0) Ignoring MS power control level 10 (10 dBm) => 9 (12 dBm): ms-pwr-lvl[curr 9, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
+(bts=0,trx=0,ts=0,ss=0) Keeping MS power at control level 10 (10 dBm): ms-pwr-lvl[curr 10, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
+(bts=0,trx=0,ts=0,ss=0) Ignoring MS power control level 10 (10 dBm) => 9 (12 dBm): ms-pwr-lvl[curr 9, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
+(bts=0,trx=0,ts=0,ss=0) Keeping MS power at control level 10 (10 dBm): ms-pwr-lvl[curr 10, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
+(bts=0,trx=0,ts=0,ss=0) Ignoring MS power control level 10 (10 dBm) => 9 (12 dBm): ms-pwr-lvl[curr 9, max 2], RSSI[curr -78, avg -78, thresh -80..-73] dBm, C/I[curr 14, avg 14, thresh 12..16] dB
diff --git a/tests/power/ms_power_loop_test.ok b/tests/power/ms_power_loop_test.ok
index 4dc01e4..2bb780d 100644
--- a/tests/power/ms_power_loop_test.ok
+++ b/tests/power/ms_power_loop_test.ok
@@ -166,14 +166,14 @@
 	MS current power 13 -> 14 (expected 14)
 
 Starting test case 'test_good_threshold_convergence'
-lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 1 (expected 1)
-	MS current power 9 -> 9 (expected 9)
-lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 1 (expected 1)
+lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 0 (expected 1)
+	MS current power 9 -> 10 (expected 10)
+lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 0 (expected 0)
 	MS current power 10 -> 10 (expected 10)
-lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 1 (expected 1)
-	MS current power 9 -> 9 (expected 9)
-lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 1 (expected 1)
+lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 0 (expected 0)
+	MS current power 9 -> 10 (expected 10)
+lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 0 (expected 0)
 	MS current power 10 -> 10 (expected 10)
-lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 1 (expected 1)
-	MS current power 9 -> 9 (expected 9)
+lchan_ms_pwr_ctrl(RxLvl=-78 dBm) returns 0 (expected 0)
+	MS current power 9 -> 10 (expected 10)
 Power loop test OK

-- 
To view, visit https://gerrit.osmocom.org/c/osmo-bts/+/25503
To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings

Gerrit-Project: osmo-bts
Gerrit-Branch: master
Gerrit-Change-Id: Ib33a94eba92004f9327c91e020a506dfa9a957c3
Gerrit-Change-Number: 25503
Gerrit-PatchSet: 1
Gerrit-Owner: pespin <pespin at sysmocom.de>
Gerrit-MessageType: newchange
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osmocom.org/pipermail/gerrit-log/attachments/20210917/b9c5460e/attachment.htm>


More information about the gerrit-log mailing list