Home RF over Glass Testing

RFoG Testing & Results / Excentis (Belgium)

Currently a lot of informations can be found regarding RFoG systems (burst mode). This informations are mostly related to vendor specific models that differs from vendor to vendor even if based on a Industry Standard. Regarding Experiences “on field” or extended “Lab tests” that provides datas about “long term” and “intensive use” of such systems, it’s hard to find respective documentation or informations. Until now, no practical oriented measurements were available.

To ensure “fully reliability” to such Tests it was necessary to identify a “Testing Institute” with highest competences in Docsis Technology. Company “Excentis” in Ghent (Belgium) is specialized in testing of CMTS for “Cable Labs Europe” and also RFoG-Systems for interested vendors.

Main objective was to test the “INTEROPERABILITY” of RFoG in a DOCSIS 3.0 environment. Specially interested to test our own

“OBI safe” solution (continuos mode), we were also interested to know, how a RFoG System (burst mode) will react in different situations.

The biggest challenge in RFoG (burst mode) is the risk of OBI (Optical Beat Interference), but not only. OBI is given by a combination of Structure oriented and Transmission oriented issues as it is for RFoG (burst mode) solutions. Within this solution all optical lines are “fix and shared” by an optical splitter. For Downstream this is not an issue but for the Upstream lines this means a “fix and shared” connection of 32 or 64 lines (Buidlings or Subscribers) to 1 single Receiver. OBI is caused when at least 2 Lasers (with same wavelength) were activated at the same time. This happens when the “signal transmission” controlled by CMTS is not more “synchronic” as it is when used Docsis 3.0 or more specific the “channel bonding” functionalities. Such “Interferences” in Upstream results as “lost of Docsis Packets”. Even the RFoG Standard (SCTE 174 2010) clearly defines this exigencies, RFoG (burst mode) vendors, relies on CMTS Manufacturers and their possibility to provide a “time scheduling functionality” to “synchronize bursty traffic”, limiting CMTS-Capacity and “Troughput”, reducing “channel bonding” to 1 channel . At the moment only 1 CMTS Manufacturer has such a functionality.

The main idea was to investigate the influence of different parameters and setting on data traffic by analyzing “lost of Docsis Packets” measuring the Codeword Error Rate (CER).

Like above mentioned, not only “Optical Interferences” is an issue but also “RF-Interferences”, generated by Turn-On and Turn-Off Characteristics (burst mode) of the RFoG Optical Network Unit. So far a second goal of the tests was to define any potential risks coming from such “burst mode” Technology.

EMC SA provided, installed and fine-tuned the optical system. Excentis tested two different RFoG systems from EMC Electronic Media Communication SA in a (Euro)DOCSIS environment and under different conditions. Both solutions “burst mode” and the “continuous mode” provided by EMC SA were tested with the same conditions and testing Parameters. Testings included:

· 95’000 measurings

· 57 different configurations

· 5 days intensive testing

Tested variables and DOCSIS Parameters

Using multiple upstream channels the following upstream related parameters/variables had to been adjusted and that have a certain effect on the system:

- Data modulation

- Bandwidth (~ Symbol Rate)

- Data load

- Forward Error Correction settings

- Burst preamble length

- Time

- Laser input power

The parameters that were used to characterize the system are:


- CCR and CER

- TCP Average Throughput

- Voice quality

OBI safe by EMC - Continuous Mode System

· Works fine using the worst case parameter set discovered in the Burst Mode System tests. The Continuous Mode System did not suffer from OBI during the tests.

· The Continuous Mode System has a significantly higher performance and stability compared to the Burst Mode System.

· The system shows a dynamic range of at least 6 dB when tested using a worst case parameter set over different RFoG transmitter input powers.

· The average TCP throughput using the Continuous Mode System is a lot more stable seen over the different modems (lasers).

· The average TCP throughput using the Continuous Mode System is significant higher compared to the Burst Mode System

RFoG - Burst Mode System

OBI Issus

· Works fine when using synchronous operation (single upstream channel (Euro)DOCSIS 2.0 mechanism).

· The system shows significant issues when using the (Euro)DOCSIS 3.0 channel bonding mechanism.

· With an optical spectrum analyzer it was clear that the different lasers were shifting significantly in wavelength while they were transmitting.

· The different wavelengths overlap, causing OBI.

· As soon as two or four upstream channels are bonded, the CER and CCR raise significantly due to OBI.

· As soon as two or four upstream channels are bonded the DOCSIS SNR drops significantly.

· Each laser shows a different CER and CCR due to its wavelength position and the CMTS (Euro)DOCSIS packet scheduling.

· The CER variations per laser are quite limited.

· SNR variations are seen over an 18 dB range.

Some statistics

· 40% of all test cases in the 4 upstream bonding configuration have
over 10% errored codewords.

· 27% of all test cases using upstream bandwidth 6.4MHz have
over 10% errored codewords.

· 26% of all test cases using upstream data modulation QAM64 have
over 10% errored codewords.

· 44% of all test cases having a load of 80% have over 10% errored codewords
while only 2% of all test cases having a load of 25% have over 10% errored codewords.

Average throughput

· The average throughput using the Burst Mode System is significant lower compared to the Continuous Mode System.

· Using QPSK modulation with 2 bonded channels results in quite low CER; the channel capacity is however rather low (20 Mbit/sec raw capacity compared to 120 Mbit/sec raw capacity using QAM64 and 4 channels)


· Calls in a worst case scenario clearly show high amounts of packet loss; even over 10%. According to voice quality scores the call was of quite poor quality. From a human ear perspective however the call was better than expected from the voice quality score. This is probably related to how the voice packet loss was spread over time.

Turn-On and Turn-Off

· No issues were observed with the Turn-On and Turn-Off characteristic of the system under test while testing different preamble lengths.

About Excentis

"Providing excellence in telecom and ICT services".
That is the core of Excentis' business. Headquartered in Ghent, Belgium, Excentis is a leading and solid provider of highly specialized testing, consultancy and training services for access network technologies (cable, xDSL, wireless, etc) and related services (e.g. VoIP and iDTV). As a trusted partner, Excentis works closely with operators and service providers worldwide to exhaustively test their equipment, from individual devices to the most complex end-to-end systems and services. Excentis acts as the independent, neutral and only certification testing lab on behalf of European cable operators. Born out of this hands-on testing expertise, the company also provides consultancy services technology selection, design and architecture evaluation, security audits, due diligence support, etc., offers standard and customized training services. Vast experience and thorough expertise of the most diverse technologies and services, combined with state-of-the-art test labs make Excentis a unique center of excellence.


About EMC Electronic Media Communication SA

A Swiss manufacturer and Pioneer within Signal transmission solutions für CATV. Main business focus is to provide system solutions to enhance a Docsis based “step by step” evolution of a HFC Network in combination with Fiber. With more then 18 Years FTTx-experience, specific Know How and own developed solutions, EMC offers a wide range of standard, solid and innovative Systems, such as „OBI safe“ (continuous mode), installed „thousands of times” and successfully established because of it’s reliability.