International Civil Aviation Organization
ACP-WG S/5
WP-06 14/7/14
WORKING PAPER
AERONAUTICAL COMMUNICATIONS PANEL (ACP)
5th_MEETING OF THE WORKING GROUP S (SURFACE)
Agenda Item
XX
Packet Error Rate Evaluation method
(Proposed by Shoichi Hanatani / Ichiro Murata)
SUMMARY
This document provides information about how Packet Error Rate (PER) has
been evaluated as the primary index for radio link quality instead of Bit Error
Rate (BER). Also it provides some detail information for measurement method
of PER, which has been adopted in WiMAX Forum,
1.
INTRODUCTION
In WP04 of Web-Meeting#5, BER estimation by packet error measurement was introduced.
The WG Meeting agreed to adopt this PER evaluation. This WP provides the actual method
of PER measurement in accordance with MRCT; Mobile Radio Conformance Tests WMFT25-002-R010v06 by WiMAX Forum. These detail descriptions, which are described in
APPENDIX are supplied to Technical Manual.
2.
2.1
PER MEASURMENT AND EVALUATION METHOD
Relation between BER and PER
A real communications system is considered equivalent to the ideal system if they have the same PER.
The equivalency means that the real system has the same packet throughput as the reference ideal system
having the known BER=10-6.
(5 pages)
ACP-WG-S/5 WPxx_Error measurement.doc
ACP-WG-S/5 / WP-xx
-2-
The PER is determined exclusively by the BER and the number of bits in the packet’s data payload and is
not dependent by how the data is encoded or what happens during the transmit-receive process. Therefore
the relationship between PER and BER is given by:
PER =1− (1− BER)n
Equation 1
The PER measurement and evaluation method is described in the APPENDIX in detail.
3.
ACTION BY THE MEETING
ACP WG-S is invited to consider following actions.
1) Confirm the PER index for a basis to extrapolate BER to evaluate maximum
BER in AeroMACS SARPs validation.
2) This methodology will be described in Technical Manual.
-3APPENDIX
PER MEASURMENT AND EVALUATION METHOD
1) Relation between BER and PER
The relationship between the BER and PER is valid for an ideal communication system that transmits
data over binary symmetric channel with uncorrelated noise. This relationship is not valid for correlated
noise channels and is dependent on the specific receiver implementation. The length of the packet in bits
is n and the bit error probability for the channel is BP. Any packet that has one bit or more in error is
discarded and the probability Pn of getting a packet in error is given by:
Pn = 1 - (1- BP)n
Equation 1
Pn is the packet error rate of n byte packet case, and BP is the bit error rate of one packet. .
A real communications system is considered equivalent to the ideal system if they have the same PER.
The equivalency means that the real system has the same packet throughput as the reference ideal system
having the known BER=10-6.
The PER is determined exclusively by the BER and the number of bits in the packet’s data payload and is
not dependent by how the data is encoded or what happens during the transmit-receive process. Therefore
the relationship between PER and BER is given by:
PER =1− (1− BER)n
Equation 2
2) Measurement Packet size and error count
To measure and evaluate BER by PER, n; packet size, should be defined first. The packet size is adopted
as follows.
For the BS receiver sensitivity, the packet length is chosen to be, after encoding, a single FEC block with
a maximum data size allowed by the CTC sub-channel concatenation rule, i.e., 60/54 bytes depending on
the particular MCS level. Each data packet consists of a 6-byte generic MAC header at the beginning and
a 4-byte CRC- 32 at the end, leaving the remaining as payload bits. Random payload bits are used. The
CRC is calculated based on MAC header and the payload.
PDU size for BS receiver sensitivity and PER corresponding to BER=10-6 are shown in Table 1.
-4-
ACP-WG-S/5 / WP-xx
Table 1 PDU packet size for BS receiver sensitivity and PER(BER=10-6)
Channel BW
MCS
Packet size
Packet size
(MHz)
(PDU Size)
(PDU size)
(bytes)
(bits)
5
QPSK rate-1/2
60
480
QPSK rate-3/4
54
432
16QAM rate-1/2 60
480
16QAM rate-3/4 54
432
Source: MRCT: WMF-T25-002-R010v06 Table 202
PER (BER=10-6)
0.048%
0.043%
0.048%
0.043%
For the MS receiver sensitivity, the packet length is chosen to be 540 bytes.,
PDU size for MS receiver sensitivity and PER corresponding to BER=10-6 are shown in Table 2.
Table 2 PDU packet size for MS receiver sensitivity and PER(BER=10-6)
Channel BW
MCS
Packet size
Packet size
(MHz)
(PDU Size)
(PDU size)
(bytes)
(bits)
5
QPSK rate-1/2
540
4320
QPSK rate-3/4
540
4320
16QAM rate-1/2 540
4320
16QAM rate-3/4 540
4320
Source: MRCT: WMF-T25-002-R010v06 Table 36.
PER (BER=10-6)
0.43%
0.43%
0.43%
0.43%
The PER is defined as the limit of the ratio between the number of the packets in error (Nape) to the
number of the packets transmitted (Not) when Npt goes to infinity as shown in Equation 3.
PER = Npe / Npt
Equation 3
However, PER cannot be measured in a finite time measurement. Instead we define experiments that,
once validated, will give a defined level of confidence that the PER is less than the requirement.
This is defined by the level of confidence α, the value of the PER, the total number of N packets
transmitted and the maximum number M of packets that can be in error for the experiment to be
considered valid. M is determined from N, PER and α that the probability for a system that has a PER
higher than the nominal one to pass the test is less than 1- α as shown in Equation 4.
The number M is the largest integer that satisfies the inequality:
K=M
Σ
K=1
N
k
P nk (1 - Pn) N-k ≦１ - α
Equation 4
For each modulation and coding alternative, the test is performed for the trial packet according to Table 2
and run until the number of frames specified in Table 3 has been transmitted. The number of packets in
error should be less than the limit indicated in the last column of the table.
-5-
For the BS receiver sensitivity measurement, if the level of confidence is 95%, N=60000, and Pn=
0.048%, for the instance of a packet size of 60 bytes, the BER would be less than 10-6 with confidence
greater than 95%, when M is equal or less than 20.
That is, if M is equal or less than 20 by these conditions, the BER would be less than 10-6 with a 95%,
level of confidence. It would meet the receiver sensitivity measurement criteria. If the packet size is 54
byte and N = 68000, M is equal or less than 21, the possibility of BER, that is less than 10-6 would be
more than 95%.
The measurement number of frames is chosen as a realistic measuring system. In the measurement
system, one target packet is included for each transmit slot, which has a 5 ms cycle. When 60000 packets
would be measured, the duration should be 300 seconds or 5 minutes.
The column; # of error packets shows M, derived from Equation 4. For each MCS below, the test is
evaluated at 95% confidence level.
In the case of QPSK 1/2 for UL, at least 60000 frames are measured, and the number of the error packets
should be equal or less than 20, for proving its receiver sensitivity.
Table 3. Parameters for Single-antenna BS Receiver Sensitivity (PUSC)
Channel BW
MCS
PDU Size
# of frames PER
(MHz)
(bytes) n
N
(BER=10-6)
5
QPSK rate-1/2
60
60000
0.048%
QPSK rate-3/4
54
68000
0.043%
16QAM rate-1/2
60
60000
0.048%
16QAM rate-3/4
54
68000
0.043%
Source: MRCT: WMF-T25-002-R010v06 Tabel.202
# of error
packets M
20
21(Note1)
20
21 (Note1)
(Note1): The original value defined in WMF-T25-002-R010v06 is 20, however we conclude the correct
value shall be 21 after careful review and verification.
For the MS case, the level of confidence is 50% according to MRCT, and N=30000, Pn= 0.43% in case
of a packet size of 30000 bytes, M is equal or less than 129 should be adopted as shown in Table 4.
Table 4. Parameters for Single-antenna MS Receiver Sensitivity (PUSC)
Channel BW
MCS
PDU Size
# of frames PER
(MHz)
(bytes) n
N
(BER=10-6)
5
QPSK rate-1/2
540
30000
0.43%
QPSK rate-3/4
540
30000
0.43%
16QAM rate-1/2
540
30000
0.43%
16QAM rate-3/4
540
30000
0.43%
Source: MRCT: WMF-T25-002-R010v06 Tabel.36
# of error
packets M
129
129
129
129