1 Assignment AMAP False Detection Analysis Document Number: IEEE S802.16m-10/0314 Date Submitted: Source: Yi Hujun Yin Intel Corporation Re:Category: AWD-Comments / Area: Chapter (DL physical structure) “Comments on AWD DL physical structure” Venue:IEEE Session#66, Orlando Purpose: Discussion and Adoption by TGm Notice: This document does not represent the agreed views of the IEEE Working Group or any of its subgroups. It represents only the views of the participants listed in the “Source(s)” field above. It is offered as a basis for discussion. It is not binding on the contributor(s), who reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE Patent Policy: The contributor is familiar with the IEEE-SA Patent Policy and Procedures: and. Further information is located at and.

Assignment AMAP Design Decisions in TGm Various AMAPs are separately coded for different users. MLRU is the minimum resource unit for assignment AMAP. MLRU is formed from distributed LRUs in the time first manner. Two MCS levels can be used for assignment AMAPs in an AMAP region. The two MCS levels can be either QPSK (½, ¼), or QPSK(½, 1/8), depending on configuration. ¼ TBCC is used to code assignment AMAP Non-user specific information indicates the number of assignment AMAP in an assignment AMAP group. CRC in Assignment AMAP IE is masked by station ID. –Blind detection using station ID is needed.

False Positive Detection False positive detection happens when CRC check passes by mistake for a detection trial. This is caused by certain decoding errors such that error bits in the AMAP IE and CRC are aligned to allow CRC check to pass. Such decoding errors can be caused by –Channel impairment or RF impairment. This factor cannot be avoided in any blind detection scheme. –Rate matching or repetition mismatch. While present in LTE, this cause of false positive is eliminated in the current structure using only 1 AMAP IE size. Because encoded assignment AMAP is not scrambled with any user specific sequence, each MS can decode assignment AMAP for other MS if the channel is ideal. However CRC check would fail because it’s scrambled by STID. With this in mind, false detection is less likely to happen for MS with good geometry and more likely to happens for cell edge users. AMAP false positive detection can interrupt both DL an UL traffic, especially for UL persistent scheduling. LTE has many contributions on the analysis of the false activation frequency of semi-persistent scheduling (SPS).

A-A-MAP Decoding Errors From SINR distribution of EMD baseline cell, 50% of users can receive QPSK ½ correctly. Assume the rest of the users can receive QPSK 1/8 correctly. 50% of users with good geometry can decode all A-A-MAPs correctly so there is no false detection for those users. The rest of the users can decode A-A-MAPs of QPSK 1/8 correctly but fails to decode A-A-MAPs of QPSK ½ correctly. Assume half of the A-A-MAPs use QPSK ½ and the other half uses QPSK 1/8. Given 32 A-A-MAPs per subframe, on average there is 8 decoding errors per users.

False Positive Probability Assume the worst case scenarios in terms of false detection: –20 MHz bandwidth TDD (4:4). –Fully loaded VoIP traffic. –32 A-A-MAPs and detection trials –8 A-A-MAP decoding errors per MS on average Suppose an MS tries to decode all A-A-MAPs in each DL subframe, a false positive detection can happen every 2 16 /(8*4*200)=10.24 seconds. Because there are 16 A-A-MAP IE types and each type has equal probability of false detection, false detection duration of a UL IE (UL basic and UL persistent IEs) is * 8 = seconds.

False Detection Impact When A-A-MAP false detection generates a false DL A-MAP IE, MS would try to decode a false DL burst and send a Nack in UL at a arbitrary HARQ feedback channel. –If the UL HARQ feedback channel is used by another MS, the worst case is that the Nack transmission wipes out an Ack transmission of the other MS. This makes BS performs a redundant retransmission. The system impact of this effect is small. When A-A-MAP false detection generates a false UL A-MAP IE, MS can transmit an invalid data burst which can collide with other UL transmission. The collision can cause a failure of an HARQ process. –Further analysis is needed to see how much impact the false detection can affect the failure rate of the system HARQ process. The failure rate of HARQ process is designed to be between to

UL HARQ Failure Rate Due to False Detection (1) The following scenario is used to analyze the impact of false detection on UL HARQ processes. –20 MHz 4:4 TDD –Full VoIP capacity traffic –32 A-A-MAPs and detection trials per DL subframe From the evaluation results submitted to IMT-Advanced, 16m VoIP capacity is about 80 users/effective MHz per link. In the considered scenario, it means 800 active VoIP users in the system. Given 800 VoIP users, one UL IE false detection can happen once every 81.92/800 = 102.4ms. Given 800 VoIP users, there are 400 UL HARQ processes in 20 ms (one VoIP packet every 20 ms per user and 50% VoIP activity factor) or 2000 UL HARQ processes in 100 ms.

UL HARQ Failure Rate Due to False Detection (2) Assume that a false UL A-A-MAP detection causes –A invalid UL data burst transmission. Subsequent transmission due to synchronous HARQ and persistent scheduling is stopped due to detection of invalid DL HARQ feedback. –Collision with another one UL burst (a little optimistic) –HARQ process associated the collided UL burst fails due to the collision (a little pessimistic) The HARQ process failure rate due to UL A-A-MAP false detection is therefore 1/2000 = 5 ∙ With the target failure rate of HARQ to be between to 10 -4, the HARQ failure rate caused by UL A-A-MAP false detection should be reduced.

UL A-A-MAP IE Reserved Bits IE TypeIE parametersReserved bits with 4 Tx ant. Reserved bits with 2 Tx ant. UL Basic IESU-PMI=0, CSM=0, MEF=0 67 SU-PMI=0, CSM=0, MEF=1 56 SU-PMI=0, CSM=1 23 SU-PMI=125 UL Persistent IETNS=Mt (SU- MIMO) 33 TNS >Mt (MU- MIMO) 22

A-A-MAP IE Sanity Check Based on Reserved Bits As shown in the previous slide, UL A-A-MAPs have different numbers of reserved bits depending on the IE type and some parameters in the IE. MS should perform IE sanity check after A-A-MAPs are successfully decoded. If the decoded A-A-MAP is from a false detection, MS may detect invalid reserved bit value and identify false detection. Given n reserved bits, sanity check can reduce false detection probability by 1/2 n if A-A-MAP IE randomization is applied.

A-A-MAP IE Sanity Check Based on Resource Allocation When false detection happens, the resource allocation may be invalid such as allocation overlapping with UL control region and allocation occupying different types of resources. For VoIP based traffic, MS expects small resource allocation. For example, a 12.2 kbps VoIP burst contains no more than 40 bytes. –Given the burst size signaling design in 16m, the burst size index (I B ) is the sum of allocation size index (I A ), defined as the product of LRU number and rank, and five bit MCS (I MCS ) in assignment AMAP, i.e., I B =I A +I MCS. –For no more than 40 bytes burst sizes, I B is from 1 to 16. Therefore I A can be from 1 to 16 as well. For each given I A, I MCS can be from 0 to 16-I A. The total number of valid combination of burst allocation size, MCS, and rank is (16+1)*16/2=136. –There are 8+5+2=15 bits in UL persistent scheduling IE used for resource allocation, MCS, and rank respectively. –By sanity check, MS can reduce the probability of false detection by a factor of 136/2 15 =

Sleep Mode Implementation of power saving modes like sleep or idle modes effectively reduces the need to decode assignment AMAP constantly and therefore reduces the probability of false detection as well. For VoIP traffic, there is one packet every 20ms. MS only needs to wake up one subframe every 4 frames to decode assignment AMAP for allocation or de-allocation of persistent scheduling. The probability of false detection can therefore be reduced by a factor of 1/16.