|
|
Broadband Bob Report
|
|
|
Introduction This workshop was sponsored by the Society of Cable Telecommunications Engineers(SCTE) jointly with IEEE Communication Society on September 25-27, 1996 in Tucson, Arizona. The main theme of the workshop is " High Integrity Fiber-Coax Networks." The purpose of the workshop is to bring together technical experts to discuss the issues and determine economic ways of optimizing HFC network design to ensure that these networks become the broadband two-way access networks of choice. Key topics discussed covered a wide spectrum, including spectral integrity and protection, survival network configurations, powering, encryption, security and telecommunications services network integrity specifications as well as an update on the progress by various standards bodies is a total of 22 papers presentations. Overall, the workshop was well attended (140), well run and worth the effort because of the great opportunity to network with others in the cable industry and the chance to gain some insights from the technical and marketing presentations. Although there were enough technical contents in the papers presented, lack of original material was very apparent and much of the data used for availability and reliability analysis was drawn from the same old well. Key Issues Details a) Standards and Technical Requirements - New near term services are voice, internet access, broadcast digital video over semi-integrated platform - The late 97 will see a transition to integrated broadband services platform, i.e. additional new services - The ultimate services platform, which does not exist should be modular, re configurable, scalable, and economical He also suggested that there are five deadly assumptions: - every thing has to work in 6 MHz channelization - everything must use IP layer 3 - voice services must look and feel like incumbent offering - everything must be built to last 20 years - we have to wait for and adhere to existing standards - ATM Forum Residential Broadband Working group - IEEE 802.14 working group - Digital Audio Video Council (DAVIC) - Multimedia Cable Network System Partner Ltd.(MCNS) - Society of Cable Telecommunications Engineers (SCTE) high speed data working group The ATM forum is making progress towards resolving the VPI and VCI multiplexing and concentration, traffic management support, UNI signaling termination, service model definition, voice telephony over ATM support and interface to 802.14. The work plan calls for first internal draft to be issued on December 96. As of Sept. 1996, the following are the major items for IEEE 802.14 consensus: - 64 QAM on downstream - QPSK/16 QAM for upstream transmitters - QPSK for upstream receivers - ATM cells as one of the basic transport units - Upstream symbol rates (Kps): 256, 512, 1024, 2048, 4096 - DAVIC/DVB Forward Error Correction (FEC) for downstream - ATM cell transport mandatory in cable modems and headend - Variable length packet support optional in cable modem and headend DAVIC is further ahead in either, data solution was secondary to the digital video to the set-top-box, now cable modem specification in progress. MCNS represents rapid path towards cable modem interoperability for US cable industry. As for the SCTE high speed data working group, it has just started work with no work plan available at this time. b) Transmission Spectral Integrity c) Transmission Availability and Security -Even the base 480 homes passed LEC and 2000 HP MSO fiber nodes provide sufficient upstream throughput to most data service requirements at projected subscription rates. - Node splitting to 120(LEC), and 500 (MSO) home nodes provides an enormous increase in the upstream data throughput per user, without requiring any new technology - Adjunct Mini fiber nodes increase upstream data throughput even further - Using the Advanced Digital Access Protocol (ADAPt) which is a contribution of Lucent to IEEE 802.14, further enhances the data services capabilities of HFC network. He concluded the following: - Typical HFC networks are capable of reliability performance comparable to Bellcore's 53 minute/year standards if backup power can be provided - Backup power well in excess of 12 hours is needed to meet 53 minutes/year. - Due to the longer hold times of video, however, 53 minutes of downtime with video will be more noticeable to customers than 53 minutes of telephone downtime - While the Bellcore standards relate to a single customer view point, some consideration should also be given to the total number of customers impacted per oulittlecyberes - Redundant portions of the system (such as route-diverse fiber) typically don't contribute to downtime. It is important to notice that this paper did not take the drop into consideration for reliability calculation and the number of coax connectors in the model seemed very small compared to the number of taps. - must guard against not only the probable, but also possible threats - must assume component failures and design flaws from low level up to the system level. He warns that security is like a chain and will likely break at its weakest link. The adversary is smarter, better educated, more motivated, and better funded, and has a much longer time to break security than the designer have to design the system. The paper presented two examples of security: Public key-based key management via factory crypto-initializing for HFC telephony and public key-based key management via user crypto- initialization for cable modems The paper concluded that: 1) the modulation choice is not a key differentiator; it is recommended to use the highest order that the channel can support. 2) Access scheme is the differentiator with S-CDMA as very strong candidate to be used in the HFC. d) Specific HFC Broadband Services requirements The paper concludes with that applications enabled by today's HFC networks allow operators to: differentiate their offering, reduce churn, derive additional revenue sources, increase brand equity, and leapfrog the competition. - understand bi-directional transmission environment - find ways to test and quantify cable system performance - provide test and operational guidance to cable operators who will be using HP modems - assist HP's cable modem design team. They ran their tests on the following systems: - an HP in house lab system, - a cable system with isolated fiber node which has no customers connected to it, - a cable system with full bi-directional node, - a cable system with bi-directional trunk/feeder The in-house test bed was 750MHz HFC, while the rest were only 550MHz. The tests in the downstream include: headend C/N, node C/N. end of line EOL C/N, downstream reflections, group delays, downstream clipping. In the upstream they tried to quantify statistically random signal and transient ingress events, also to measure the effect of manipulating upstream environment such as filters, node size, equalizers signal levels and 5 MHz high pass filters. The downstream clipping test results indicated that vollittlecybere peaks occur in groups spaced 167 ns, C/N increases linearly with modulation; CTB and CSO change slowly and smoothly. CSO of-69 dBc, CTB<-70 dBc. Also the results indicated that end of line errors are not significantly higher than just fiber node, which may imply that clips are likely primary error causes. In the upstream the following was observed as characteristics of "Typical" transient pulse/burst: - distinct clipping limit in one polarity - 10-40 ns rise and fall times - pulse width of 10-50 ns, internal burst repetition rate 3-10 MHz - no consistency in burst length: single pulse to > 16 ms observed - use of scope plus analyzer allowed both detailed and envelope observation of pulses. The author indicated that there still some remaining upstream work to be done such as: - correlate pulse amplitude with bit errors - statistical studies of ingress, both carrier and impulse - measure effects of drop equalizers, filters, node size, operating levels - develop method for easily quantifying system channel quality, both upstream and downstream e) Traffic Prediction, Network Sizing, Modulation Techniques For the analysis, it was assumed that the user is using a 90 MHz Pentium with Windows 95 OS surfing the web during the busy hour (20-35% active users). A "typical" surfing sequence was generated accessing several head-end cached web pages with significant graphical content. The paper concluded the following: - current hardware 'cable modem' access architecture will provide throughput for ~200 users that will greatly exceed customer expectations. - Current node size of 500 HP nodes are quite reasonable. - New hardware is approaching a scalable target architecture - Cable access network provides superior burst, broadcast and multicast capabilities. Peter concludes that OSSs will play a major part in the management of the network and its operations supplemented by a complete plant performance monitoring capability. He ends with an interesting "reliability versus cost" chart which he called "Improvement Opportunities Summary." The chart measured in terms of $ per home passed that suggests a compromise that is based on the nature of the business opportunity. Parallel Sessions HFC Powering Consideration for Enhancing Availability Out of 40 slides presentation, 30 slides discussed elementary availability and reliability concepts. The rest of the charts compared centralized powering versus distributed powering with no clear winner. The paper concluded that: Powering architecture and design critically impact network reliability. Noise/Interference Control Albert Kim of Rogers Cable presented this paper which covered some of Rogers Cablesystems Two-way network issues such as, network segmentation size, network controllable elements, Rogers integrated operational support system. Rogers operational experience indicated the following: - proposed reverse performance levels are achievable - two-way sweep equipment are absolutely mandatory - status monitoring/remote capability are required - methodical, step by step vetting required; "do the job right the first time" - Comprehensive systematic approach is required. |