Throughput knockdown effect on ePMP 1000 Force 200 when testing with ITU-T Y.1564

Hi,

we recently had a throughput issue on ePMP 1000 Force 200 configured in PTP TDD mode.

The link is running at maximum MCS profile (15/15) with good RSSI (-51/-51) and SNR (42/42).

We are using Flexible D/U ratio, 5 ms frame size and no DFS.

Running the embedded Wireless Link Test we get the following results:

128 byte packet size -> 19 Mbps downlink / 16 Mbps uplink

800 byte packet size -> 120 Mbps downlink / 89 Mbps uplink

1500 byte packet size -> 189 Mbps downlink / 126 Mbps uplink

Running a similar test with ITU-T Y.1564 test suite, using EMIX:abceg frame size mix and pushing 96 Mbps in the link, the final results reports a throughput of 18 Mbps.

I monitored the SNMP port counters on LAN and WAN interface during the ITU-T test and found that there is a kind of knock down effect on radio transmission when trying to push this specific traffic mix + amount in the equipment.

Is this an expected behaviour?

Is this caused by the low throughput performance on small frame size (EMIX a, b & c) and by the buffer congestion and overflow due to being not able to forward that big amount of small size packets?

I need to explain this "poor result" to the End User.

I made some calculations.

Consider that ITU-T Y.1564 defines EMIX as

a: 64 byte size packet

b: 128 byte size packet

c: 256 byte size packet

e: 1024 byte size packet

g: 1518 byte size packet

and it requests to transmit the select packet sizes in sequence.

I.e. the default and recommanded EMIX:abceg means 64>128>256>1024>1518>64>128>256>1024>1518>.....

The test just select the right inter-packet pause for transmitting the right TX bitrate.

Making some very basic approx. math I found that:

a+b+c frame size packets are 15% of the aggregate (a+b+c+e+g) 96Mbps TX rate, i.e. c.a. 14,3 Mbps

As we know from Wireless Link Test with 128 byte packet size we reach a max throuhgput of c.a. 16 Mbps

Then this mix of a+b+c frame size packets inside the aggregate a+b+c+e+g flow are using almost the full throughput resources.

The remaining e+g frame size packets contribute to completely congesting the tx radio buffers and this causes the knock down effect with massive Capacity Drop events.

This is a very stressful test for the radio link...

my friend what is firs you link based  of linkplannet send pictures,  and exacly what is you problems is slow  PTP mode conection datta transfer modes size packets? 

Dude seriously you really gotta work With your vocabulary, it kinda gross the way you’re trying to write in English.



Best Regards.

Kevin Cedeno F.
NPO and EPC Wireless Engineer
EWPE - Ericsson LTE Certified Engineer
MWFE - Motorola Wireless Field Engineer
Landline: +593 97-952-2253
Mobil WS : +593 98-896-8995
Landline EUA: +1 (518) 712-9466
Ecuador - Guayaquil
United States - California
Unless you try to do something beyond what you have already mastered, you will never grow.

Well it could be. But you have to understand what Half and Full Duplex Means.

Bet you get low T-Put capacity because of your config, take a deep look into that. If you need any help
Do not hesitate in contact me .



Best Regards.

Kevin Cedeno F.
NPO and EPC Wireless Engineer
EWPE - Ericsson LTE Certified Engineer
MWFE - Motorola Wireless Field Engineer
Landline: +593 97-952-2253
Mobil WS : +593 98-896-8995
Landline EUA: +1 (518) 712-9466
Ecuador - Guayaquil
United States - California
Unless you try to do something beyond what you have already mastered, you will never grow.