Vo IP Testing SCTE New Jersey Chapter 9 13 07

Vo. IP Testing SCTE New Jersey Chapter 9/13/07

My Business Card Larry Jump Regional Sales Engineer Sunrise Telecom 814. 692. 4294 ljump@sunrisetelecom. com Confidential & Proprietary 2

Today’s Agenda § DOCSIS Troubleshooting Pyramid § RF Impairments § DOCSIS Problems § Vo. IP Impairments Confidential & Proprietary 3

Troubleshooting Methodology Troubleshooting any DOCSIS network begins with a BOTTOM UP approach. Today, typically 80% of the problems are RF and 20% are IP/DOCSIS related. Vo. IP Lives Here Communications between CMs, MTAs, subscribers & IP servers Communications between CMTS & CMs only What we fix today Confidential & Proprietary 4

Building Blocks of Troubleshooting Evolution of RF Troubleshooting • RF impairments have typically been viewed as the root cause of all network problems in cable networks • Since two-way data services have been introduced, cable operators have evolved the RF plant to higher and higher standards • Although RF impairments are still readily present in cable networks, it is important to acknowledge that other impairments exist which must be addressed DOCSIS & IP Protocols > 20% of Impairments • • DHCP / TFTP / To. D / DNS / CMS / etc. servers Modem and MTA configuration files, CMTS configs Confidential & Proprietary 5

Build a Solid RF Foundation for the Pyramid 1. Sweep/balance for alignment 2. Return path node certification and maintenance 3. Home installation quality of workmanship and materials 4. Conform to DOCSIS Specifications for signal quality Confidential & Proprietary 6

Forward Sweep Headend or Hub Site Forward Signal Path H R L § Ensures Unity Gain in the Forward path. Unity Gain minimizes distortions on all forward signals. § Ensures that entire frequency band arrives at the customer’s receive site with equality. § Ensure good sweep and 95% of the other problems go away! Confidential & Proprietary 7

Return Sweep Return Signal Path H R L § Ensures Unity Gain in the Return path. Unity Gain minimizes distortions on all return signals. § Ensures that entire frequency band from all subscribers arrives at the headend or hub site with equality. § Ensure good sweep and 95% of the other problems go away! Confidential & Proprietary 8

Downstream Measurements Confidential & Proprietary 9

Modulation Error Ratio § MER is used as the single figure of merit for DVBC standards. § It includes distortions such as CCN, CSO, CTB, laser compression, etc…. The sum of all evils. § A 256 QAM picture tiles at 28 d. B § A minimally good MER is 31 d. B for 256 QAM at the back of the customer’s set. H Confidential & Proprietary 10

Modulation Error Ratio (MER) § Modulation Error Ratio (MER) is a measure of the phase and voltage variation. Symbol Error And expressed mathematically by: • Finally, aggregate MER is just the sum of multiple MER measurements: Each point represents one symbol (2 -bits) in QPSK of data or one phase position. The distance from the circle is the error. Confidential & Proprietary 11

DOCSIS Compliant Downstream § Analog Measurements § CNR § CSO § CTB ( ≥ 35 d. B per DOCSIS spec) ( ≥ 41 d. B per DOCSIS spec) § Analog (Digital) Measurements § BER (post-FEC 10 -8 or less per DOCSIS spec) § Modulation Error Ratio (MER) § Constellation Analysis § Digital Channel Power 1 error in 100 Million bits Confidential & Proprietary 12

Typical BER/MER Requirements BER 64 QAM MER 10 -10 >35 10 -8 27 -34 10 -6 23 -26 10 -5 <23 256 QAM MER Quality >35 Excellent 31 -34 Good 28 -30 Marginal <28 Fail NOTE: Set-top boxes can tolerate some Post FEC errors, but cable modems cannot. Confidential & Proprietary 13

Downstream Digital Measurements Confidential & Proprietary 14

Constellation Analysis Noise CW Interference Phase Noise I/Q Gain Error I/Q Phase Error Compression Confidential & Proprietary 15

Equalizer Stress § Digital receivers use adaptive equalizers to negate the effects of § § signals arriving other than the desired signal. Signals can arrive ahead of or after the desired signal. In a cable system, the majority of signals are reflections and microreflections that arrive after the desired signal. Cable modems and digital set top boxes must be able to handle pre and post (DELAYED) signals at levels defined by DVB standards. If the equalizer is pushed beyond those limits, errors will occur. By using the Velocity of Propagation, the distance to the source of the reflection can sometimes be located. If the reflections occur before the next upstream amplifier, they are simply amplified and passed downstream thereby eliminating the ability to perform fault detection based on reflection time. Equalizer stress is used more as a figure of merit for the margin available to the set top box or cable modem. Confidential & Proprietary 16

Cable Modem Downstream Confidential & Proprietary 17

Downstream Frequency Response DOCSIS Specifies <. 5 db peak to valley per MHz Confidential & Proprietary 18

Intermittents § While you can’t measure a problem that isn’t happening, there may be clues § Any reading that is not normal for your system may be suspect § Example -Lower than normal MER/BER § Erodes headroom and error margin – any degradation will cause issue § Higher than usual ingress/noise § Check Ingress in Reverse AND Forward Path § Any CPD § CPD levels are often variable – may be minor now – major tomorrow § Use Statistical Measurements to monitor over time Confidential & Proprietary 19

Cable Modem Downstream § Stats Mode Measurements Graphed over time § MER and Pre and Post BER measured over time Confidential & Proprietary 20

Upstream Certification Confidential & Proprietary 21

DOCSIS Compliant Upstream § Linear Impairments such as: • Micro-reflections -10 d. Bc @ <= 0. 5 µsec (per DOCSIS spec. ) -20 d. Bc @ <= 1. 0 µsec -30 d. Bc @ > 1. 0 µsec • Amplitude ripple (0. 5 d. B/MHz per DOCSIS spec. ) • Group Delay (200 ns/MHz per DOCSIS spec. ) § Non-linear Impairments such as: • Common Path Distortion (CPD) • Return Laser Clipping § Transient Impairments such as: • Ingress & Impulse Noise (CNR > 25 d. B per DOCSIS) § MER / Pre & Post-BER Channel Characterization Confidential & Proprietary 22

Effects of Over-Driving a Laser Confidential & Proprietary 23

CPD – Note the 6 MHz Marker Delta! Because the channels in the forward system are 6 MHz apart, the sum and difference frequencies occur at 6 MHz intervals as well. 6 MHz Confidential & Proprietary 24

Upstream Characterization SPECTRUM ANALYZER with QAM DEMOD 16 -QAM TRANSMITTER Confidential & Proprietary 25

Upstream Laser Compression Constellation Confidential & Proprietary 26

Micro-Reflections §Micro-reflections are an indication of a mismatch in the network. § A problem because the mismatches reflect the incident signal back towards the source causing standing waves (or ripples) in the amplitude of the frequency response where these 2 signals collide Confidential & Proprietary 27

Group Delay § Group delay occurs at the roll-off points of the diplex filters and its effect gets worse with more filters in the cascade. (remember there are 2 for every active) § Group delay affects MER; to achieve minimum MER levels for 16 -QAM group delay must be reduced to a certain level Confidential & Proprietary 28

Which means? As different frequencies pass through a Cable System, some will move faster than others— Group Delay 5 MHz 10 MHz t 10 MHz 15 MHz 20 MHz 25 MHz 30 MHz 35 MHz 40 MHz SYSTEM Filters & Traps 15 MHz 20 MHz 25 MHz 10 MHz SYSTEM Filters & Traps 15 MHz 20 MHz 25 MHz 30 MHz 35 MHz 40 MHz T I M E Confidential & Proprietary 29

Group Delay § DOCSIS specifies 200 n. Sec/MHz • But <70 n. Sec/ MHz is recommended for Vo. IP in 16 -QAM modulation 240. 6 ns/MHz at 35 MHz!!! Confidential & Proprietary 30

16 -QAM with Group Delay Confidential & Proprietary 31

Time Lapsed Analysis of US Channel MER Pre-FEC Post-FEC Confidential & Proprietary 32

Return Path Monitoring Early Fault Detection and Faster Node characterization for Vo. IP and other return services Confidential & Proprietary 33

Moving Up the Troubleshooting Pyramid § Now we can move from RF diagnostics to DOCSIS CALL SIGNALING & VOICE TRAFFIC DQo. S–SERVICE FLOWS And DOCSIS Measurements Confidential & Proprietary 34

Cable Modem Connect Confidential & Proprietary 35

Cable Modem Detail Confidential & Proprietary 36

Signaling and Service Flows § There are 2 types of traffic in a Vo. IP call, signaling and § § § the actual voice transmissions The signaling sets up the call path and tears it down after the call is completed. Signaling also provides dial tone, ring, and ring back. Service flows are simply a system of prioritizing digital transmissions. Some service flows take priority over others. Service flows only exist between the CMTS and the CM. Many systems use best effort data transmission for signaling and Qo. S for voice. A Qo. S service flow always gives voice traffic priority and allocates separate bandwidth for that traffic. Confidential & Proprietary 37

Call Signaling and Service Flows MTA CM CMTS CM MTA Call Signaling (best effort) Service Flow Add RTP Call Flow (Qo. S) Call Signaling (best effort) Service Flow Del Confidential & Proprietary 38

Some Common “DOCSIS” Call Preventers? Call Signaling Fails to Go Through • “Best Effort Service” competes with other traffic • Usually TCP/IP signaling will go through, but customer may not wait for dial tone or digits • CMS receives excessively delayed digits from DOCSIS contention region – REQuest – Grant period used by other best effort services such as Vonage, gamaing, etc. • Remedy Establish dedicated Qo. S for Call Signaling with (10 kbps) per e. MTA, drawback is uses US BW Call Disconnects After Ring • e. MTA and CMTS unable to establish DQo. S – Bad e. MTA, not Packet. Cable certified or bad Packet. Cable certificate in e. MTA – e. MTA CODEC or configuration file mis-configured – CMTS out of Service Flows – Failure to delete inactive SIDs Confidential & Proprietary 39

Test Best Effort & Service Flow Channels STEP 1 STEP 2 Test the IP Best Effort Service Flow - Packet Loss, Latency & Jitter - Best Effort Verification for Call Signaling § Test the Upstream & Downstream Voice Quality – Vo. IP MOS & R-Factor tests – Using DOCSIS Service Flow (Qo. S) Confidential & Proprietary 40

Testing Best Effort Call Signaling § Test the performance of your systems “Best Effort” services § Good for testing network performance for call signaling § Will your e. MTA establish communication with the CMS? § Measures §Packet loss §Latency §Jitter Courtesy Sunrise Telecom Broadband Note: Does not use Qo. S (DOCSIS Service Flow) Confidential & Proprietary 41

Moving Up the Troubleshooting Pyramid § Now we can move from DOCSIS to IP CALL SIGNALING & VOICE TRAFFIC DQo. S–SERVICE FLOWS And DOCSIS Measurements Confidential & Proprietary 42

Building Blocks of Troubleshooting §IP traffic is quickly becoming an “impairment” when it reaches excessive levels, overwhelming the Cable Modem Termination System (CMTS) § Excessive CMTS utilization causes Vo. IP subscribers to experience poor call quality, data users experience slow downloads/uploads, and in worst case scenarios the data network comes to a halt! § Conventional RF monitoring tools cannot identify IP traffic utilization § Capacity planning tools become essential § How do we manage the traffic on the network and our subscribers before the CMTS becomes oversubscribed? Confidential & Proprietary 43

Throughput Test Confidential & Proprietary 44

Two Categories of VOIP Testing 1. Network Components Latency Jitter Lost Packets 2. Voice Quality Mean Opinion Score R Factor Confidential & Proprietary 45

Latency Measurement Latency is simply the transit time between one network element to another End to end delay LATENCY = TIME Toll Quality <150 m. Sec Confidential & Proprietary 46

Types of Delay or Latency § Transportation Delay § Delay caused by the packet to get through the network components such as routers and gateways. This is controlled by the system architecture. § Propagation Delay § Simply the time it from one place to the other. § Packetization Delay § This is the time it takes to build the voice packets in the MTA § Jitter Buffer Delay § Delay caused by the jitter buffer Confidential & Proprietary 47

Latency Causes § Voice must be digitized, optionally compressed, processed for echo canceling, and packetized. § Voice packets may take multiple hops. § Result is that voice over IP networks has more delay than traditional circuit switched approaches. § Solutions § Run with very small packetization period (10 ms). § Minimize processing delays in system components. § Minimize number of hops from source to destination. § Give voice packets priority. Confidential & Proprietary 48

How is Latency Measured? Network Entity Directing packets through a UGS pipe (service flow) provides a good platform for testing latency to the network side of the CMTS PING PING Qo. S Service Flow PING DOCSIS Analyzer Confidential & Proprietary 49

Vo. IP Issue - Jitter § § Jitter is simply variations in Latency Delay in routers vary with current traffic load. Voice packets can take different routes. Net result is a variability in delay, this is called jitter. In other words the packets don’t always arrive in the order they were sent. § Voice is a real time communication so the packets must play out in the order they were transmitted § Solution. § Jitter buffer at the playout side. § Received packets are placed here first before playout. § Builds in a standard delay to allow packets to arrive, get buffered up, then played out. § User hears no gaps between delayed packets. § Give voice packets priority. Confidential & Proprietary 50

Jitter = Change in Latency Packet Sequence In 1 2 3 4 5 Packet Sequence Out UGS Service Flow 3 2 4 1 5 Transmitted packets evenly spaced in time Received packets unevenly spaced in time Packets are experiencing changes in transit time (Latency) Delta Latency = Jitter Most gateways use buffering to compensate for jitter, but buffering can contribute to latency Confidential & Proprietary 51

Incoming voice packets 3 2 4 1 5 Jitter Buffer § Voice packets have sequence numbers so a buffer can reconstruct the incoming traffic § Buffer plays out packets in sequence order § Jitter Buffers cause delay 1 2 3 4 5 To Listener Confidential & Proprietary 52

What’s Good? What’s bad? Latency - Numbers vary but…. . Preferred <60 m. Sec Maximum 150 m. Sec Very annoying about 250 m. Sec Jitter Preferred equal to or <20 m. Sec Maximum should be <30 m. Sec Confidential & Proprietary 53

Vo. IP Issue - Lost Packets § There is no time for retransmission! § Packet loss can occur due to traffic or physical layer problems § Solutions – PLC Packet Loss Concealment § If single packet is lost, replay the last packet because the human ear is very forgiving. § After one replay, play white noise. Confidential & Proprietary 54

VOIP & Lost Packets VOIP is very sensitive to lost packets A 1% packet loss will cause the call to break up A 3% packet loss will drop the call completely Some telcos spec 0. 5% packet loss or better for High Speed data/voice circuits Confidential & Proprietary 55

Voice Quality Tests MOS – Mean Opinion Score Actual listeners grade performance on a scale of 1 -5 A more modern method uses a MOS server and send voice data packets to the server for analysis. The server also sends voice data back to the instrument which also performs analysis Using this method, both upstream and downstream are analyzed separately. R Factor is essentially the same thing only rated on a different scale based on 100 instead of 5 Confidential & Proprietary 56

Factors Affecting Voice Quality § Latency, Jitter and Lost Packets § Noise § Delay § Echo § Intelligibility Confidential & Proprietary 57

Cable Modem Vo. IP Measurements § Cable Modem or Ethernet Vo. IP Tests § Downstream § Upstream § CMTS Round Trip § Select Test Length Confidential & Proprietary 58

Knowledge Check What is the primary source of impairments in a DOCSIS network NCS signaling is sent in a DOCSIS 1. 1 service flow for guaranteed Qo. S. True or False DOCSIS specifies that the CM / e. MTA achieves a Post-FEC BER to be less than or equal to what Confidential & Proprietary 59

Thank You! Confidential & Proprietary 60