Presentation time & sequence numbers
Richard Maiden
RF Accuracy
What do we care about?
2
eNB
RoE
Radio
Nasty
stuff
RoE
Radio over
ethernet
What do we care about?
3
eNB
RoE
Radio
Nasty
stuff
FIFO
Nice
predictable
stuff RF
RoE
What do we care about?
4
eNB
RoE
Radio
Nasty
stuff
FIFO
D D
wrA rdA
Nice
predictable
stuff RF
RoE
What do we care about?
5
eNB
RoE
Radio
Nasty
stuff
FIFO
D D
wrA rdA
Nice
predictable
stuff RF
RoE
Delay magic
Elastic
Clock magic
What do we care about?
6
eNB
RoE
Radio
Nasty
stuff
FIFO
D D
wrA rdA
Nice
predictable
stuff RF
RoE
Delay magic
Elastic
Clock magic
Don’t care sample
Clock and delay magic
7
Out of scope
Many, many possible schemes
– SyncE
– 1588v2
– GPS
– PWM
– RTT measurements
– etc.
Many, many concerns
– Frame delay variation
– One time / regular
– Closed/open loop
– PLL, OVCXO, fiber length, #hops
– Jitter, Wander
8
Classes and granularity
Timing accuracy
From 802.1CM (Apr 6th)
19 April 2016 IEEE 1904 Access Networks Working Group, City, Country 9
Nice to have – MIMO
Must – Carrier aggregation
Must – Carrier aggregation
Must – LTE TDD
Jitter scenario
CPRI focus here (#1 use case)
CPRI basic frame is 1/3.84MHz
– 260.41666666’ns
Our timeStamp is 0.25ns
19 April 2016 IEEE 1904 Access Networks Working Group, City, Country 10
Scenario 1
19 April 2016 IEEE 1904 Access Networks Working Group, City, Country 11
basic frame count / 3.84
Round(Prefect presentation time / 0.25ns)
+/-83ps
Quantized
Scenario 2
19 April 2016 IEEE 1904 Access Networks Working Group, City, Country 12
+/-83ps is pretty small
But we either transmit a sample or we don’t. A drifting clock would be worse
19 April 2016 IEEE 1904 Access Networks Working Group, City, Country 13
Desired egress Gap in egress Overlap egress
1 sample
@ 10MHz LTE
is 65ns
Alternatives…
Do nothing– its out of scope
Force / suggest nice increments
– Anything divisible
by 3 works ok
Increase accuracy
– Reduces error but
not slip?
Variable step size
– 1/3.84 rather
than 0.25ns
19 April 2016 IEEE 1904 Access Networks Working Group, City, Country 14
Motion #x
making the motion
Seconded by
Technical motion (>=2/3)
Yes: -, no: -, abstain -
19 April 2016 IEEE 1904 Access Networks Working Group, City, Country 15