28
doc.: IEEE 802.11-14/0890r3 Submission July 2014 Youhan Kim, Qualcomm I Slide 1 WLAN-3GPP Interworking Metric Date: 2014-07-17 N am e C om pany A ddress Phone em ail Y ouhan K im Qualcom m youhank@ qca.qualcomm.com M enzo W entink Qualcom m mwentink@ qti.qualcomm.com CarlosA ldana Qualcom m caldana@ qca.qualcomm.com Authors:

Doc.: IEEE 802.11-14/0890r3 Submission July 2014 Youhan Kim, Qualcomm Inc.Slide 1 WLAN-3GPP Interworking Metric Date: 2014-07-17 Authors:

Embed Size (px)

Citation preview

doc.: IEEE 802.11-14/0890r3

Submission

July 2014

Youhan Kim, Qualcomm Inc.Slide 1

WLAN-3GPP Interworking Metric

Date: 2014-07-17

Name Company Address Phone email Youhan Kim Qualcomm [email protected]

Menzo Wentink Qualcomm [email protected]

Carlos Aldana Qualcomm [email protected]

Authors:

doc.: IEEE 802.11-14/0890r3

Submission

Revision History

July 2014

Youhan Kim, Qualcomm Inc.Slide 2

Revision Note

R0 Initial

R1 Fixed document number typo.Minor editorial updates (e.g. fixed figures moving around).Changed references from REVmc D3.0 to IEEE 802.11-2012

R2 Changed reference 14/0921r1 to 14/0921r2

R3 Changed reference 14/0921r2 to 14/0921r3

doc.: IEEE 802.11-14/0890r3

Submission

Background

• 3GPP TSG RAN WG2 (RAN2) is developing a mechanism for interworking (IW) between 3GPP RATs (UMTS and LTE) and WLAN– 3GPP had sent a liaison statement to IEEE 802.11 in April 2014

(IEEE 802.11-14/0519r0)

– IEEE 802.11 responded as in IEEE 802.11-14/0658r6• See next slide

• Discouraged use of RCPI and RSNI

July 2014

Youhan Kim, Qualcomm Inc.Slide 3

doc.: IEEE 802.11-14/0890r3

Submission

Previous Liaisons

July 2014

Youhan Kim, Qualcomm Inc.Slide 4

Questions in liaison statement from 3GPP(IEEE 802.11-14/0519r0)

Response from IEEE 802.11 to 3GPP(IEEE 802.11-14/0658r6)

Question 1: Does IEEE 802.11 WG consider WLAN RCPI a suitable metric of WLAN signal strength such that it can be compared to thresholds as in the above described mechanism?

We consider the RCPI value as defined in IEEE 802.11™-2012 a metric for signal strength.

Question 2: Does IEEE 802.11 WG consider WLAN RSNI a suitable metric of WLAN signal quality such that it can be compared to thresholds as in the above described mechanism?

We consider the RSNI value as defined in IEEE 802.11™-2012 a metric for signal quality in downlink direction.

doc.: IEEE 802.11-14/0890r3

Submission

Previous Liaisons (Cont’d)

July 2014

Youhan Kim, Qualcomm Inc.Slide 5

• 3GPP TSG RAN WG2 (RAN2) is developing a mechanism for inter-working between 3GPP RATs (UMTS and LTE) and WLAN– 3GPP had sent a liaison statement to IEEE 802.11 [1] in April

2014, to which the IEEE 802.11 responded as in [2]

Questions in liaison statement from 3GPP(IEEE 802.11-14/0519r0)

Response from IEEE 802.11 to 3GPP(IEEE 802.11-14/0658r6)

Question 3: Does IEEE 802.11 WG consider any other WLAN signal metric more suitable for the above described mechanism?

Understanding that the objective of the mechanism is to select the network that provides the best match to the QoS and/or throughput requirements of the system, the consideration of RNSI/RCPI is not sufficient on its own to efficiently estimate the available throughput and QoS that will be experienced in the IEEE 802.11 WLAN. Other metrics should be taken into account, especially channel bandwidth, operating band, number of spatial streams, BSS load, and WAN metrics, see also the attached Table 1. Comparing only the RSNI/RCPI, as is, to thresholds presents some risks of poor decisions. Ideally, a single parameter, such as estimated available throughput, which combines all of the above parameters, would be determined inside of the WLAN modem and then delivered to the upper layers.

doc.: IEEE 802.11-14/0890r3

Submission

Further Background

• 3GPP RAN2 is still using RCPI and RSNI as the PHY metrics for WLAN-3GPP IW

– ftp://ftp.3gpp.org/tsg_ran/WG2_RL2/TSGR2_86/Docs/R2-142955.zip– ftp://ftp.3gpp.org/TSG_RAN/TSG_RAN/TSGR_64/Docs/RP-140556.zip– ftp://ftp.3gpp.org/TSG_RAN/TSG_RAN/TSGR_64/Docs/RP-141010.zip– ftp://ftp.3gpp.org/TSG_RAN/TSG_RAN/TSGR_64/Docs/RP-141011.zip

• 3GPP RAN2 had requested 3GPP SA2 to include RCPI and RSNI in the requirements documents

– RAN2 is the technical group– SA2 is the Architecture requirements group

• During discussion in 3GPP SA2, additional issues were found for RCPI and RSNI

– ftp://ftp.3gpp.org/tsg_sa/WG2_Arch/TSGS2_104_Dublin/Docs/S2-142942.zip

• This contribution describes additional issues with using RCPI and RSNI for WLAN-3GPP IW, and proposes way forward for WLAN-3GPP IW

July 2014

Youhan Kim, Qualcomm Inc.Slide 6

doc.: IEEE 802.11-14/0890r3

Submission

RSNI

July 2014

Youhan Kim, Qualcomm Inc.Slide 7

doc.: IEEE 802.11-14/0890r3

Submission

• IEEE 802.11-2012

Definition of RSNI in IEEE 802.11

• P.7

• P.20

July 2014

Youhan Kim, Qualcomm Inc.Slide 8

doc.: IEEE 802.11-14/0890r3

Submission

Definition of RSNI in IEEE 802.11 (Cont’d)• IEEE 802.11-2012

• 10.11.9.4. Noise Histogram report: P1069

July 2014

Youhan Kim, Qualcomm Inc.Slide 9

doc.: IEEE 802.11-14/0890r3

Submission

Example• STA1 is DUT

• BSS1 and BSS2 have partially overlapping channel– Also, AP2 close to STA1

– But BSS2 is lightly loaded. So, channel is clear for BSS1 most of the time.• Good candidate for 3GPPWLAN roaming for STA1

• Assume noise floor of -96 dBm/20 MHz at STA1 RX

July 2014

Youhan Kim, Qualcomm Inc.Slide 10

doc.: IEEE 802.11-14/0890r3

Submission

Example

July 2014

Youhan Kim, Qualcomm Inc.Slide 11

doc.: IEEE 802.11-14/0890r3

Submission

Example• Case 1

July 2014

Youhan Kim, Qualcomm Inc.Slide 12

doc.: IEEE 802.11-14/0890r3

Submission

• Case 2Example

July 2014

Youhan Kim, Qualcomm Inc.Slide 13

doc.: IEEE 802.11-14/0890r3

Submission

Example

• Before receiving the first green packet, STA1 RX likely listening to the entire 80 MHz– Don’t know the BW of the next packet coming in.– Need to perform CCA on 80 MHz

• Then, should be ANPI for the first green packet be over 20 MHz or 80 MHz?– Of course, a ‘logical’ answer is that it should be over 20 MHz (-96 dBm), not 80 MHz (-90 dBm)– But the point here is that the standard does not define this clearly – yet another example of potential pitfall.

• Case 3

July 2014

Youhan Kim, Qualcomm Inc.Slide 14

doc.: IEEE 802.11-14/0890r3

Submission

Other Notes

• What if DUT has multiple RX antennas?– How is ANPI defined?

• Average over RX chains? Summed over RX chains?

– Not defined clearly in the IEEE standard Interop issue expected w/o further clarification

• RSNI measurement is optional

• RSNI is NOT RSRQ in 3GPP– RSRQ = Reference Signal Received Quality

• Roughly speaking, SINR

– RSNI does not measure the interference ‘present’ in the packet– Rather, the interference is measured during ‘idle’ time some time before the

packet arrives• That interference may or may not be present in the packets the DUT is receiving

July 2014

Youhan Kim, Qualcomm Inc.Slide 15

doc.: IEEE 802.11-14/0890r3

Submission

Summary – RSNI

• RSNI is currently broken– RSNI definition fails (numerically cannot be computed) in some

cases

– Many ambiguities exists

• WLAN-3GPP interworking should not be based on a metric which is broken

• RSNI is not RSRQ

July 2014

Youhan Kim, Qualcomm Inc.Slide 16

doc.: IEEE 802.11-14/0890r3

Submission

RCPI AND RSSI

July 2014

Youhan Kim, Qualcomm Inc.Slide 17

doc.: IEEE 802.11-14/0890r3

Submission

RCPI vs. RSSI

• RCPI and RSSI are similar in nature– RCPI is the power measured during the data portion of the packet– RSSI is the power measured during the preamble portion of the packet

• HT/VHT preamble in case of HT/VHT PPDUs– Signal power between preamble and data are the same– There is minor difference in noise bandwidth

• RCPI assumes noise equivalent BW 1.1 times greater than the channel BW• No noise equivalent BW specified for RSSI• Difference should be small

• Measurement accuracy– Dominated by analog gain inaccuracy, not digital power measurement

inaccuracy• RCPI and RSSI should be able to achieve similar accuracy

• Accuracy requirement in IEEE 802.11– RCPI has a ±5 dB (95% confidence interval) accuracy requirement, while

RSSI does not.• But as mentioned above, there is no reason why RSSI cannot achieve similar

accuracy as RCPI

July 2014

Youhan Kim, Qualcomm Inc.Slide 18

doc.: IEEE 802.11-14/0890r3

Submission

RSSI is Mandatory

• RCPI is optional

• RSSI, on the other hand, is mandatory– Used for CCA

• CCA is key functionality of CSMA, which all WLAN devices have to implement

– IEEE 802.11-2012: P1811

July 2014

Youhan Kim, Qualcomm Inc.Slide 19

doc.: IEEE 802.11-14/0890r3

Submission

RSSI Unit

• RSSI unit in IEEE 802.11 is a ‘relative’ dB, not an ‘absolute’ dBm

• But since RSSI is used for CCA– CCA is in absolute dBm

– Thus, RSSI in absolute dBm is implicitly available

July 2014

Youhan Kim, Qualcomm Inc.Slide 20

doc.: IEEE 802.11-14/0890r3

Submission

Other Ambiguities

• Multiple RX chains– RCPI for DSSS/CCK and NON_HT OFDM PPDUs does not

specify how to deal w/ multiple RX chains• E.g. Average over RX chains? Summed over RX chains?

• HT/VHT PPDUs specify that the RCPI is averaged over RX chains

– RSSI also does not specify how to deal w/ multiple RX chains

July 2014

Youhan Kim, Qualcomm Inc.Slide 21

doc.: IEEE 802.11-14/0890r3

Submission

Summary – RSSI vs. RCPI

• RSSI and RCPI convey essentially the same information– RCPI is optional

– RSSI is mandatory

• Both RSSI and RCPI has some ambiguities to be clarified

July 2014

Youhan Kim, Qualcomm Inc.Slide 22

doc.: IEEE 802.11-14/0890r3

Submission

BEACON VS. DATA

July 2014

Youhan Kim, Qualcomm Inc.Slide 23

doc.: IEEE 802.11-14/0890r3

Submission

What Packet Should be Used?

• Before WLAN has associated w/ an AP– There is no traffic. Only Beacon RSSI is available.

• When performing active scan, Probe Response RSSI would be available instead of Beacon RSSI

• But typically, Probe Response and Beacon are both sent in the lowest supported rate

• After WLAN has associated w/ an AP– If device is using 3GPP link as the main data pipe, then WLAN will not have

much traffic• WLAN will likely be in DTIM mode to save power

– Again, only beacon RSSI is available

• Even if data traffic is available– Devices often change TX power as a function of MCS

• Several dB of TX power difference between the lowest rate and the highest rate is not uncommon

– Hence, for the same path loss, data packet RSSI could vary considerably just because of TX power

• Receiver has no information on what the TX power was– Hence, data packet RSSI is not suitable for link quality assessment

July 2014

Youhan Kim, Qualcomm Inc.Slide 24

doc.: IEEE 802.11-14/0890r3

Submission

Beacon RSSI

• IEEE 802.11-2012– P354

July 2014

Youhan Kim, Qualcomm Inc.Slide 25

doc.: IEEE 802.11-14/0890r3

Submission

Summary – Beacon vs. Data RSSI

• Beacon RSSI is always available– Before association

– After association, w/ and w/o traffic

• Data packet RSSI is not as reliable for link quality accessment– Function of TX power, which is implementation specific (i.e. can

change drastically)

July 2014

Youhan Kim, Qualcomm Inc.Slide 26

doc.: IEEE 802.11-14/0890r3

Submission

SUGGESTED WAY FORWARD

July 2014

Youhan Kim, Qualcomm Inc.Slide 27

doc.: IEEE 802.11-14/0890r3

Submission

Suggested Way Forward

• Communicate the following to 3GPP– Do not use RSNI for IW

• Fixing RSNI would take some effort

• Besides, RSNI does not necessarily reflect interference present in the packet. Hence, benefit for WLAN-3GPP IW is not clear

– Replace RCPI w/ Beacon RSSI• Need some clarifications in the IEEE 802.11

– Please see IEEE 802.11-14/0921r3 for details of the proposed changes

• For longer term, IEEE 802.11 should work on enhancing metrics for WLAN-3GPP interworking

July 2014

Youhan Kim, Qualcomm Inc.Slide 28