COMMITTEE T1 TELECOMMUNICATIONS Working Group T1E1.4 (DSL Access) Costa Mesa, California; March 8-12, 1999 T1E1.4/99-132 CONTRIBUTION TITLE: SOURCE*: PROJECT: AM Ingress on xdsl Loops Nortel Networks T1E1.4, ADSL Standards Project ABSTRACT This contribution presents the findings from activities investigating the levels of ingress present on residential loops due to AM radio transmitters. Additionally, analysis of potential impacts of such levels on Splitterless ADSL performance is presented. NOTICE This contribution has been prepared to assist Standards Committee T1 Telecommunications. This document is offered to the Committee as a basis for discussion and is not a binding proposal on Nortel Networks. The proposed requirements are subject to change in form and numerical value after more study. Specifically, the right to add to, or amend, the statements contained herein is reserved by Nortel Networks. * CONTACT: Edward J. Eckert; email: ejeckert@nortelnetworks.com; Tel: (919) 992-3787; Fax: (919) 992-7892 AUTHORS: M. Drew; email: mdrew@ nortelnetworks.com; Tel: (613)765-4036; Fax: (613) 765-4523 K. Harris, email: kharris@nortelnetworks.com; Tel: (613)765-2923; Fax: (613) 765-4523
1. Introduction The demand for high-speed data access by residential users is speeding the deployment of a variety of Digital Subscriber Line systems (xdsl) into many territories. To ensure market acceptance and customer satisfaction, a substantial amount of effort has gone into the specification of loops and impairments for use during product qualification. The current G.992.1, G.992.2 and G.996.1 specifications include loop topologies, crosstalk characteristics, impulse noise and POTS activity. Interference due to AM radio ingress is not addressed in these specifications. 2. Background All North American urban centers are served by local AM radio transmitters which are located and engineered in such a way as to serve large numbers of potential listeners. The number of transmitters serving any city tends to increase with population density due to the ability to costeffectively reach a large audience and competitive pressures. Similarly, early xdsl deployment will be most prevalent in urban areas with higher population densities allowing access providers to spread the capital and infrastructure costs over a large numbers of potential customers. Energy from off-air transmitters, such as AM radio, can couple onto twisted pair loops and result in interference at an xdsl receiver. There are a number of factors influencing the differential ingress level on a loop due to a specific AM transmitter. They include the transmitter characteristics such as transmitted power, antenna pattern orientation and proximity. The characteristics of the customer loop also impact the resultant interference. Relevant characteristics include, aerial/buried, length, orientation, drop type and balance. The existence of RFI ingress as a significant impairment has been recognized in the VDSL standards work [1] and it is included as a noise contributor. 3. Ingress Tests Measurements were performed on residential loops located in two Canadian cities. The study included differential ingress levels measured on 42 loops of varying make-ups. Test sites were geographically distributed throughout the cities and include both aerial and buried plant in 60% and 40% proportions respectively. Measurements were performed at the Network Interface Device (NID) with the in-premise cabling disconnected. Differential levels were measured across a 100Ω differential termination without a common-mode connection. 1
4. Measurement Results Interference due to individual AM transmitters was clearly evident at almost every home with wide variations observed from home to home. Figure 1 depicts the distribution of levels when the single highest interferer is considered at each home. The worst case interference at the median home is 53 dbm while 10% of homes would experience a worst case interferer of 37 dbm or higher. Maximum AM RFI Maximum Level (dbm) 0-10 -20-30 -40-50 -60-70 -80-90 -100 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of Loops Figure 1 Cumulative percentage of loops with differential AM ingress levels below a maximum level Further analysis of the measurement data focussing on the two strongest interferers at each site revealed that the difference between the power levels of the strongest and second-strongest interferer was always less than 6 db, with a median difference of 1.3 db. Figure 2 depicts the maximum ingress level data from Figure 1 with the addition of a second trace corresponding to the second strongest interferer level measured at the homes. 2
Maximum Level (dbm) 0-20 -40-60 -80-100 -120 Maximum AM RFI Two Strongest Interferers 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of Loops Most Powerful Interferor Second Most Powerful Figure 2 Cumulative percentage of loops with differential AM ingress levels below a maximum level for two most significant interferers 5. Potential Impact of AM Ingress on Splitterless ADSL Capacity The effect of AM ingress on downstream capacity when no precautions are taken to limit its influence was simulated in the following way: Frequency domain simulation using tones 36 to 127 Crosstalkers and loops as defined in standard test cases -140 dbm/hz white noise No front end filtering No TDQ No self next (echo) AM Ingressor placed at 540 khz AM Ingressor modeled as a flat white noise 10 khz wide The level of the AM Ingressor (in 10 khz bandwidth) was varied from 30 dbm to 60 dbm, and the capacity compared to that obtained with no ingressor present. The following chart shows the capacity for many of the test cases: 3
Capacity versus AM ingress level at 540 khz 100% 90% % of Capacity with no AM Ingress 80% 70% 60% 50% 40% 30% 20% TestCase 2 3 4 5 6 7 8 9 Extended 11 Extended 12 Extended 13 Extended 14 Extended 15 Extended 16 10% 0% -60-50 -40-30 AM ingress level at 540 khz in dbm Figure 3 Percent of Capacity with no AM Ingress versus AM ingress level for Selected Splittlerless ADSL test cases 4
It may be observed from the graph that common levels of ingress can have a dramatic effect on splitterless ADSL capacity. As expected those test cases with highest loss are most effected. Test cases with a large portion of lower carriers removed due to cross talk interference are also more severly impacted. The reason for the extent of lost capacity from a narrowband interferer is the noise spreading effect of the windowed DFT sidelobes. The first sidelobe is only 13 db down from the adjacent carrier and subsequent sidelobes fall off at a rate of only 6dB/octave change in frequency. Thus, if precautions are not taken to limit the amplitude of the noise source before the DFT is performed, a narrow band noise source can corrupt all bins. Figure 4 shows the signal at the input to the DFT in test case #2 with a 40 dbm AM ingressor. Figure 5 shows the spreading of the noise observed at the output of the DFT. Figure 4 Signal and Noise Profiles at input to DFT for Test Case #2 with 40 dbm AM ingress at 540 khz 5
Figure 5 Signal and Noise Profiles at output of DFT for Test Case #2 with 40 dbm AM ingress at 540 khz It should be noted that, although an in band tone at 540 khz was choosen to show the potential capacity loss due to AM ingress, out of band tones can be even more harmful if they are inadequately filtered and aliased into the middle of the downstream band. It is also important to note that the homes experiencing high ingress due to poor balance will tend to have high levels for more than one AM station. 6. Conclusions and Recommendations AM ingress is a ubiquitous form of interference. Many subscribers have a significant amount of AM ingress on their loops. There are a wide variety of solutions possible to limit the damage done by AM ingressors, and there use will greatly mitigate the capacity loss. However, failure to address the problem could give poor and erratic performance for a significant percentage of subscribers. Those most at risk are the long reach customers specifically targeted by the splitterless ADSL product. Test cases should be developed and included in the splitterless and full-rate ADSL recommendations or associated T1 pointer standard documents as well as in performance 6
benchmarking suites to help customers assess performance in the presence of this common interferor. 7. Further Work Further work is needed to produce a reasonable model of AM and HAM ingressors. The template used for VDSL may provide a suitable starting point. Capacity targets will need to be set for whatever model is applied. Contributions on this work are welcomed. References: [1] T1E1.4/98-043R6, Very-high-speed Digital Subscriber Lines, Nov. 1998 7