Institutional links

• Quick Links

  • Today's Releases
  • Decisions, Notices and Orders
  • Public Proceedings

Privacy and Security
Transparency

Report

On

Third Party ISP Access

To

Major Canadian Cable Systems

For the CRTC

 

Prepared by:

Imagineering Telecom Inc.
1090 Don Mills Road, Suite 601
Toronto, Ontario, Canada
M3C 3R6 

Date: January 2, 2002

---

Executive Summary

The purpose of this study is to provide the CRTC with technical advice to assist the Commission in establishing policy for the administration of third party access to Canadian cable television systems for data services, referred to as third party Internet access (TPIA).

Imagineering Telecom Inc. has been asked to consider and report on the issues relating to the compatibility of cable modems used by third party Internet Service Providers (ISPs¹). The mandate of the study is to provide sufficient information for the Commission staff to make a recommendation on what procedure should be followed to identify modems that are compatible with cable networks. We have, therefore, structured this report in a manner which provides an understanding of the issues and presents the research in an objective way, which we hope will lead to the desired conclusion.

We have been asked to present our research by responding to five basic questions; they are provide below in abbreviated form with a summary response:

1. Should the modems be DOCSIS compliant?

MSOs should migrate their systems to a DOCSIS compliant system, and, indeed, this upgrade to DOCSIS is already underway in all four of the major MSOs. The DOCSIS standard allows for open access cable systems and will allow straightforward implementation of TPIA. Non-DOCSIS portions of existing systems can remain in place, as long as there is a DOCSIS-based overlay facilitating TPIA.

2. Which release of DOCSIS should be permitted (1.0 or 1.1)?

Either can be permitted, though we recommend adoption of DOCSIS 1.1, but beginning with DOCSIS 1.1 hardware compliance, as the minimum standard, with a reasonable period of time for upgrade to this level, given that certification wave of DOCSIS 1.1 modems by CableLabs is now underway with completion scheduled for the end of 2001. Because DOCSIS 1.0 CMTSs and cable modems are fully compatible with DOCSIS 1.1 CMTSs and cable modems, any combination of the two in the system will operate on a best effort basis. If the cable operator chooses to implement 1.1 CMTSs, it will be able to provide QoS guarantees to DOCSIS 1.1 cable modems on the network. The natural trend will be to move to DOCSIS 1.1, to provide value added services.

3. Whose certification for DOCSIS compliance should be required?

Testing by CableLabs or a suitably qualified, independent testing facility is acceptable. Independent testing facilities offering such services do exist and could also be used, subject to a thorough qualification process. There appears to be no fundamental reason not to permit third party testing, provided that suitably qualified alternatives to CableLabs can be verified. CableLabs could continue as the certifying body by reviewing test reports prepared by independent test labs and issuing compliance certificates, as is its current practice.

4. Once a modem has received the appropriate certification, is any further testing by the cable operator necessary?

We believe that, at this stage of implementation of data over cable services and the imminent introduction of DOCSIS 1.1 capability, second level testing by the MSOs is necessary for the reasons listed below.

• Discovery of Unanticipated Problems
• Unique Hardware and Software Versions
• Software Creep
• Manufacturing Tolerances
• System Loading
• Increasing Complexity of DOCSIS standards

5. If additional testing is required what should these tests be and who should carry them out?

Currently, MSO tests typically encompass management, billing, operation, and network interoperability of the cable modem. It should be possible to phase out most of the second level testing, once DOCSIS devices become sufficiently mature, especially those complying to the DOCSIS 1.1 standard, but it is possible that some minimal amount of MSO testing will always be required, due to unique cable plant parameters, equipment and software loads. A summary of factors that, in our view, cause MSOs to perform second level testing now, and may perpetuate some level of testing on an ongoing basis, is given in the bulleted items above.

Table of Contents

Introduction 
Internet Access Over Cable

Introduction 
Cable Modems and Standards 
CableLabs Certification 
Third Party Test Laboratory Certification 
Second Level Modem Testing by MSOs 

Forced Compliance to Standards Upgrade Path 

Introduction 
Status of MSO Migration to DOCSIS 
Third Party Internet Access in Proprietary Systems 
Minimum Required Standard for MSO Cable Systems 

Positive Trends and Tools for TPIA  

Equipment Evolution 
OSS/BSS Software Developments 
Loosening of the Cable Modem Supply Chain 

Conclusions and Answers to the Commission's Questions 

Glossary of Terms 

Bibliography 

Industry interviews 
CRTC Documentation 
CableLabs Certification Documentation 
Various World Wide Web Sites 

Appendix A - Scope of Work 

Introduction

The CRTC has determined that, in the provision of Internet access and related data services, whether over the existing telephone or cable television network infrastructures, competition is beneficial to the public. To facilitate this competition, the incumbent telephone and large broadcast distribution undertakings have been directed by the Commission to allow third party Internet Service Providers access to their customer base to provide competing Internet data services over the existing cable and telephone networks. With respect to the implementation of TPIA over cable, a working group was formed to address the various practical issues related to the transport of third party ISP data over the incumbent's HFC based data network.

It was decided by the High-Speed Working Group (HSWG) that third party Internet access would be achieved using points of interconnection (POIs) at the network or IP layer, managed by the MSO. A device such as a router or Broadband Access Server² would be connected to many CMTS units in the MSO's network. TPIA customers would build a connection from their own IP network to these POI devices using standard techniques, i.e. high-speed connections provided by a common carrier (ATM, Ethernet LAN extension, or similar) or a fibre connection owned and built by the ISP. These connections to the POI have been discussed in the HSWG CISC proceedings. Using policy routing to determine traffic paths based on the source IP address of the traffic, MSOs can route IP packets through the POI to the appropriate customer network without allowing ISPs direct access to RF spectrum on the RF segment of the hybrid fibre-coax (HFC) network. CRTC/HSWG proceedings have defined the POIs and the methods by which ISPs can connect to them.

At this stage of the HSWG proceedings, there are a few key issues upon which the committee is having difficulty reaching consensus; these include:

• Adoption of open standards (DOCSIS) to facilitate TPIA
• Feasibility of TPIA over cable plants not conforming to DOCSIS (proprietary)
• Cable modem certification and testing by CableLabs and/or other testing facilities and the related issue of CM network compatibility
• Second level testing of CMs by MSOs

The Commission has decided to seek input from an independent third party on these matters, and a contract was awarded to Imagineering Telecom Inc. to study these specific issues and to provide information to the Commission to assist it in making a decision on the matter of third party access to established cable systems. The Commission's scope of work for this study is attached as Appendix A.

Internet Access Over Cable

Introduction

Cable Internet access is implemented on existing one-way cable networks using three additional components: reverse or two-way amplifiers in the distribution network, a cable modem on the customer's premises, and a cable modem termination system (CMTS) at the head-end of the cable network. Two-way amplifiers allow information to be sent back from the subscriber to the cable head-end. Cable modems attach to the coaxial line to modulate and demodulate data on the HFC network and send it to a personal computer by Ethernet or USB connection. The cable modem acts like a Layer 2 bridge from the perspective of the home user. The HFC network works like a shared contention-based network similar to Ethernet. At the head-end, the CMTS sends and receives user data on otherwise unused frequencies available on the HFC network. The CMTS accepts data coming from the subscribers over the shared cable network and sends it out to the Internet through the cable operator's backbone network using standard Internet routing protocols.

Cable Modems and Standards

In the early days of high-speed cable Internet access, there were no specific standards, and enterprising companies such as Terayon and Motorola each developed proprietary modulation schemes and hardware to transmit data over cable. These proprietary devices were rolled out rapidly by MSOs to gain market share in the broadband Internet access market. None of these proprietary systems provided for third party Internet access, however some built limited capabilities into later versions of their proprietary products to support TPIA on a small-scale. Eventually, the cable community agreed upon standards, and the possibility of open access is allowed for in these standards. The generally accepted standard for transmitting data over coaxial cable is called DOCSIS (Data Over Cable Service Interface Specifications), which was developed by CableLabs, the research consortium run by a large group of North American cable operators. The DOCSIS standard is a group of specifications describing the RF interface, OSS (operational support systems) interface, privacy components of the shared cable data network, and other fundamental parameters regarding two-way data communication over the HFC networks. Most MSO networks support this standard, and those that do not currently support it have plans to migrate to it in the near future. All of these companies now support DOCSIS standards, but some continue to manufacture and support their own proprietary systems, largely to support their existing customer base. Over 150 different models of cable modems and CMTSs are available that meet the DOCSIS 1.0 specifications.

An updated version of DOCSIS (version 1.1) with improved security and quality of service capabilities has been accepted by CableLabs. The first DOCSIS 1.1 certified cable modems and CMTSs were announced on September 29, 2001. Many new cable modems that are DOCSIS 1.0 certified are hardware compliant with the new DOCSIS 1.1 standard, so they can be software upgraded, once the combination of hardware and software has been DOCSIS 1.1 certified. These software upgrades can be done remotely by the CMTS, so modems can be upgraded in the field without user intervention.

CableLabs Certification

Currently, all testing for DOCSIS certification is performed by CableLabs. Products are tested in waves, each wave taking approximately 3-4 months to complete. A wide range of DOCSIS 1.0 cable modems and CMTSs have been certified and qualified to date; however, DOCSIS 1.1 equipment has been in ongoing testing since early 2001, and the first devices were not successfully certified until almost October 2001. To date, only two cable modems have been certified and only two CMTSs have been qualified under DOCSIS 1.1. As the hardware and software become more common, and the implementation of the underlying technology becomes better developed. LSI chipsets are being introduced that OEM manufacturers can easily build into CM devices, and, with each wave, more and more DOCSIS 1.1 products will become available.

Both the cable operators (HSCO010, HSCO047) and the open access advocates (HSCO053A) have raised complaints about the exclusive CableLabs testing. The MSOs, themselves, contend that CableLabs does not test cable modems adequately to negate the need for their own testing. Issues raised in the past have been:

• Transmit and receive power levels
• Adherence to limits set on upstream and downstream bandwidth rates
• Proper reporting of values queried by SNMP (i.e. modem management systems)

The cable operators are performing testing of cable modems with their own networks and their own back office systems to ensure interoperability and manageability. Some ISPs and their representatives have attributed the delay in rolling out third party Internet access and other advanced services on the inability of CableLabs to quickly certify devices, particularly the recent effort related to DOCSIS 1.1 (HSCO053A). Furthermore, The ISPs submit that CableLabs, being owned by the MSOs, could unduly delay the certification of devices with features that may add value to ISP offerings, and conversely, expedite the certification of devices with features that add value to MSO offerings, thus impeding the ISPs' ability to compete effectively. For these and other reasons, it has been suggested that it might be more appropriate to have a third party testing facility perform DOCSIS certification to ensure impartiality.

Third Party Test Laboratory Certification

This issue relates to the desire of ISPs to gain timely access to the cable-connected subscriber base. The current practice of requiring all cable modem manufacturers to submit to compliance testing by CableLabs exclusively is under some considerable debate. There is, however, the fundamental question of the availability of alternatives to CableLabs to carry out this testing, which is also addressed in this report. Some factors that drive modem manufacturers or, indeed, ISPs wishing to use a particular modem, to go elsewhere for testing are as follows:

• Timeliness - Compliance testing by CableLabs is done in waves, and these windows in time where testing of a particular modem can be done might not coincide with a particular manufacturer or ISP implementation schedule.
• Work volume capability - A single testing facility may become over loaded, resulting in undue delays in processing equipment.
• Perceived lack of impartiality - Since CableLabs is sponsored and partially owned by the MSOs, there may be the concern that the treatment and timeliness of the work carried out by CableLabs for third party ISPs may, to some extent, not be in favour of the ISP which requires the testing to be done.
• The cost - Cable modem certification is not inexpensive and competition in this area by private testing laboratories may be effective in reducing the cost.

It is our view that independent laboratory testing for DOCSIS compliance is workable. The process could be similar to that employed by other industries as, for example, certification of RF products for Industry Canada. Testing of a device is carried out by a recognized, independent individual or facility, which performs tests in accordance with published specifications that must be met. The test report is submitted to Industry Canada for review and issuance of a type approval certificate, if deemed compliant. Conceptually, a similar process could be used in the case of compliance testing for CMs and CMTSs.

The qualifications for an organization to perform such testing include, as a minimum, the following:

• Trusted and respected organization
• Well established
• Experienced in the field
• Competent to carry out the complex certification test program
• Sufficiently substantial to be potentially able to take on large volumes of work

Our search for organizations meeting the above requirements turned up companies that specialize in the manufacture of test equipment, as well as cable data communication and networking manufacturers. Two prominent companies that specifically offer DOCSIS compliance testing, as well as automated testing systems to facilitate such testing, are Spirent and Agilent. Compliance testing for both DOCSIS 1.0 and 1.1 are offered for both CMs and CMTSs. Spirent advertises the fact that they supplied the testing equipment used by CableLabs for its DOCSIS certification program. There are other testing facilities, e.g., The Tolly Group (http://www.tolly.com), National Technical Systems, (http://www.ntsxxcal.com/home.html), and Interworking Labs Inc. (http://www.iwl.com), which offer DOCSIS compliance testing as well. In this study, we have identified only firms offering DOCSIS compliance testing and related automated test equipment; we have not fully assessed the qualifications of these firms, as such an assessment is outside the scope of this study. However, there appear to be a number of suitable entities to choose from.

CableLabs, themselves, appear to be of this opinion as well, since they have recently issued a Request for Proposals for outsourcing such testing, though they did not accept any of the proposals. CableLabs has instead increased its testing capability for the new push to certify DOCSIS 1.1 systems. While there are numerous possible reasons for CableLabs' decision to continue to be the sole testing facility for DOCSIS certification, including purely commercial reasons, their full rationale for rejection of all proposals has not been shared with us, and we have not pursued this matter further, since it is outside the scope of this study. But the concept of third party testing appears to be an option that the cable companies controlling CableLabs might accept.

Our discussions of this topic with MSOs revealed a general willingness, in the right circumstances, to accept outside testing by suitably qualified firms, but there was also considerable reluctance as well. A point raised several times was that there is a vast amount of industry-specific knowledge and experience within CableLabs which may not be available in another organization, and going to a third party might result in reduced problem resolution capability and quality of testing. The opinion was held that, in the short term, testing by neither third parties nor by CableLabs would eliminate the MSOs' second level testing of third party ISP modems, or indeed their own.

It appears to us that the function performed by CableLabs, as the standards creation and certification body, at this stage of development of data communications over HFC networks, is valuable to the industry. However, CableLabs is not necessarily uniquely qualified to perform such testing, particularly in view of the fact that the central issue is data networking capability of the cable modems and not the equally challenging, but different, problems of cable TV plant design. Third party participation in testing, discovery of problems and conceptualizing possible solutions, may benefit from participation by parties whose central focus is data networking. Consideration of the issues surrounding the question of DOCSIS compliance testing, as discussed above, led to our forming the opinion that it is reasonable to permit other recognized testing laboratories to carry out this function. There appears to be no fundamental reason not to permit third party testing, provided that suitably qualified alternatives to CableLabs can be verified. CableLabs could continue as the certifying body by reviewing test reports prepared by independent test labs and issuing compliance certificates, as is its current practice.

Second Level Modem Testing by MSOs

Providing data over the HFC infrastructure is a very new technology, and the DOCSIS standards defining the technology are newer still, and evolving. Hence, the selection and approval of cable modems for end user connections has proven to be a topic of much debate within the High-Speed Working Group. Cable operators wish to retain the ability to specify and screen each modem make and model that will be allowed to connect to the cable network, on the basis that network incompatibility of a particular cable modem could cause network harm of a kind which would require repair beyond the act of de-activating or disconnecting the non-compliant cable modem from the network. Third party ISPs would like the ability to choose the modem for their customers and not be subjected to second level testing by MSOs, maintaining that DOCSIS certification should be sufficient to assure network compatibility.

In support of their position, submissions from MSOs describe earlier cable modem implementations that have been tested and certified to meet the CableLabs DOCSIS standards, but which, through vendor specific implementations, did not work optimally on the particular cable network in question (HSCO010). In some cases these modems did not provide accurate information for billing and network management, and in other cases had functional issues preventing access to the network for some users. MSOs currently certify cable modems for use on their networks before allowing them to be connected, even if they have already been certified to meet DOCSIS standards.

Testing of cable modems involves, among other things, analysis of the systems for management, billing and operations. MSOs emphasize their need to properly manage and query the cable modem for information. While certain information (RF power levels, for example) can only be obtained by querying the cable modem, other information (in particular, bandwidth and usage statistics) can be monitored and maintained, and indeed policed, by most modern, DOCSIS qualified, CMTS systems and provided to a billing system or mediation system for subsequent processing. Cisco, who have a significant portion of the DOCSIS 1.0 CMTS market, have implemented this functionality in their CMTS. Even if this was not the case, tools have existed for some time that could be implemented in routers to monitor bandwidth and usage on IP networks. As these tools are more integrated with cable systems, testing could be more specifically geared towards the functionality of the cable modem and the software to operate on the physical network. Bandwidth tracking for billing can be moved away from the customer premises and into the CMTS or router, where it is still distributed among many nodes, but allows for more certainty for the operator in the information provided.

Performance parameters which the MSOs feel they must test themselves are clearly those that might be affected by the specific as-built condition of the cable system in question and the specific hardware and software types, models and versions used in the system, particularly the networking, operations and back office elements related to the cable modems. There are many cable system parameters which can affect cable modem performance. But the DOCSIS standard's tests anticipate these, and the testing methodology can reasonably simulate cable system conditions in the lab, so it cannot be argued that performance related to these types of parameters must be verified in the field. However, there are, in our view, other factors which cannot be reasonably duplicated in a laboratory setting, and these lend credence to the MSOs' insistence on performing second level testing of cable modems. Examples of such factors are:

• Discovery of Unanticipated Problems
  It has been our experience that completely unforeseen interoperability problems frequently arise when a new element is introduced into a network. If a problem is anticipated, a procedure can usually be devised to test for its occurrence, but clearly, an unanticipated problem is hard to uncover in advance of equipment deployment into the network. Good engineering practice would dictate in-system tests prior to rolling out new network elements, particularly when the new elements are still in an early stage of development. Causes for this kind of behaviour are many and varied but often fall into one of the categories in the bulleted items below.

• Unique Hardware and Software Versions
  The inability of laboratory based tests to uncover problems that may surface in a particular cable system can be related to the extreme difficulty of reproducing a test bed sufficiently representative of a particular operator's cable plant and network in a lab environment. While test equipment may be able to exercise a particular modem in various respects, when similar functions are initiated by the particular versions of equipment and software existing in the field, the behaviour may be different than in the lab.

• Software Creep
  Manufacturers sometimes make minor modifications in software, and sometimes in hardware, between formal changes in version number. This is sometimes done to fix minor problems which have come to light after release of the product or in anticipation of a new feature in a subsequent release of the product. CableLabs' DOCSIS Certification Wave Guidelines state that vendors must "resubmit any certified product that is modified after certification", however minor software updates often do not go through this re-certification process, in particular when they are issued to fix bugs that are causing immediate problems with the MSO system. The cumulative effect of the introduction of several such modifications may result in undesirable behaviour in some circumstances, even though in the estimation of the manufacturer the changes did not warrant a re-certification.

• Manufacturing Tolerances
  Devices submitted for certification testing can exhibit tighter tolerances on certain parameters than units produced during subsequent, large scale production. While the specifications of units fabricated after certification testing would still be within the manufacturer's tolerances, there can be sufficient difference to cause erratic behaviour.

• System Loading
  The behaviour of a contention based MAC layer is difficult to test in the lab for heavy network loading conditions. For this reason, purchasers of new or not fully hardened (mature) network elements will frequently include in the purchase contract a condition that the product is not accepted as compliant with the specification until it has been observed to operate properly in the field in a heavily loaded network for a predefined period of time.

• Increasing Complexity of DOCSIS Standards
  The configuration and setup of various customer/operator options, preference and operation parameters involve extremely large numbers of combinations and permutations of possible settings. It is almost impossible to test a piece of equipment sufficiently to arrive at complete confidence that it will operate properly in all MSOs' cable plants, operating different equipment and software loads, notwithstanding that they may be compliant to an open interface standard. The evolving nature of cable modem technology, with DOCSIS 1.1 implementations just around the corner, adds even further complexity to the testing problem.

Because of the close relationship between the MSOs and CableLabs, it is felt by the open access advocates (HSCO053A) that the MSOs should be able to require CableLabs to modify and improve their testing to cover more of the ground that the MSOs feel they must test for themselves, including OSS and BSS parameters, and specific performance capabilities and parameters which were found to be necessary by the MSOs themselves in previous second level tests. This closing of the loop, so to speak, between the MSOs and CableLabs is indeed happening, as confirmed in our discussions with the MSOs. This feedback process is reducing the MSOs' testing load but, from our conversations with the MSOs, we perceived that there would be a general disinclination on the part of the MSOs to greatly reduce or eliminate their second level testing, if the modem's DOCSIS certification were to be done by a third party organization. Some conceded though, that when the technology is well matured - and the estimated time it would take for this to occur varied widely - second level modem testing by MSOs could be reduced, or maybe even eliminated. In the future, tests performed by CableLabs, or a third party testing body, if accepted, could become the only tests required for cable modems, though it is quite possible that this ideal may never be fully achieved.

As was stated above, cable operators do not necessarily consider a modem that has been DOCSIS certified by CableLabs to be compatible with their networks. The MSOs currently perform their own second level testing on these modems to ensure that they satisfy all of their operational and management requirements. The Commission has asked if the second level tests performed by the cable carriers primarily relate to billing issues or if they are testing the technical operation of the cable modems. In our discussions with the MSOs, there was generally no emphasis placed on any particular aspect of implementation of TPIA. Our initial expectation was that the OSS systems, and particularly the BSS systems, would be the greater source of problems. Contrary to this, the MSOs indicated that they are taking these problems in their stride. However, it must be kept in mind that, aside from some field trials, such as the one with AOL and Videotron, they have not had much experience in dealing with TPIA integration issues, particularly those that will arise when they begin to deal with the added complexities of advanced services enabled by DOCSIS 1.1.

In the short term, it is our view that MSO qualification testing of third party cable modems should be permitted for a reasonable period to allow the technology to mature and to improve the scope of tests performed by CableLabs. As for the longer term, it is not possible for us now to predict when, if ever, second level testing by MSOs will become unnecessary. In the end, it may still be necessary to permit the MSOs to conduct a minimum suite of tests in critical areas related to their specific network circumstances. In the future, the Commission could periodically review the state of technology relative to cable modems and decide whether continued MSO testing is necessary.

Forced Compliance to Standards Upgrade Path

Introduction

The Commission has the option in its proceedings to regulate the use of a particular standard, in effect, forcing cable operators to upgrade their networks to that service level. The following sections discuss some of the issues related to this possible course of action.

Status of MSO Migration to DOCSIS

Of the four largest cable carriers in Canada, two are currently providing Internet access using DOCSIS 1.0 standard equipment, and plan to upgrade to version 1.1 when hardware becomes available. All new cable modems being installed in the field are DOCSIS 1.1 hardware compliant, ready for a software upgrade to take advantage of the new features of DOCSIS 1.1. The other two use proprietary systems, but have stated their intention to migrate to the DOCSIS standard when products are available for DOCSIS 1.1, and not to provide TPIA until then (HSCO047, HSCO048). The proprietary systems in use in Canada do not easily allow for third party Internet access to take place.

As of September 29^th, 2001, two cable modems and two CMTSs have been certified for DOCSIS 1.1 compliance. The next wave of certification will be announced on December 28^th, 2001, at which time more products can be expected to be fully qualified for DOCSIS 1.1 operation.

Cable operators who are already running DOCSIS 1.0 standard networks have been installing only DOCSIS 1.1 hardware compliant cable modems since they have been available. These devices will require only a software upgrade to be fully DOCSIS 1.1 compatible, once they have passed the CableLabs certification.

DOCSIS 1.1 is a fully backward compatible standard. DOCSIS 1.0 cable modems will work with DOCSIS 1.1 CMTS devices and vice versa. When DOCSIS 1.0 cable modems are used in a DOCSIS 1.1 environment that supports quality of service, they will operate on a best effort basis. If the CMTS is performing QoS techniques with DOCSIS 1.1 modems on the same network as DOCSIS 1.0 modems, certain types of traffic originating from the DOCSIS 1.1 modem could be given reserved bandwidth, while the DOCSIS 1.0 modems would always work on a best effort basis. Having DOCSIS 1.0 cable modems share a network with DOCSIS 1.1 cable modems, even with quality of service running between the DOCSIS 1.1 enabled CMTS and cable modems, should not cause network problems but it can cause some difficulty with latency sensitive applications.

While DOCSIS 1.1 and 1.0 cable modem systems are interoperable, some of the advanced features of DOCSIS 1.1 are limited by the presence of DOCSIS 1.0 cable modems on the same network. Quality of service capabilities are implemented in DOCSIS 1.1 through the use of packet fragmentation, controlled by the CMTS. This allows increased control over jitter and delay in applications such as VoIP. Because DOCSIS 1.0 cable modems do not support packet fragmentation, a DOCSIS 1.0 modem could affect the QoS capability of a DOCSIS 1.1 modem by sending a large, non-fragmented packet over the network. This large packet would take longer to send than the small packets produced by the DOCSIS 1.1 modems, causing an unexpected delay of packets with QoS controls, and possibly causing significant problems for the application sending the packets requiring QoS. It is our understanding that some American MSOs are addressing this problem by segregating DOCSIS 1.1 and 1.0 modems on different RF carriers, thus avoiding the problem. In a system where QoS is not implemented, however, DOCSIS 1.1 and DOCSIS 1.0 cable modems can coexist without such segregation.

Both versions of the DOCSIS standard support TPIA, and most MSOs are already purchasing DOCSIS 1.1 hardware compliant modems to facilitate upgrades to the next level of service and to allow for quick implementation of TPIA. As DOCSIS 1.1 devices become available, MSOs will be able to upgrade rapidly, with the completion of the next wave of modem certification at the end of this year. Some MSOs will wait until their preferred modem manufacturer achieves certification before upgrading.

Notwithstanding the general trend in the cable industry towards DOCSIS standardization, MSOs have made significant investments in their cable systems, to provide high-speed Internet access to cable subscribers based on proprietary data networking solutions. Many MSOs built these systems before the DOCSIS standards were finalized and clearly wish to continue to use the proprietary equipment until it is reasonably depreciated. The approach that is being taken to protect this investment, while providing for DOCSIS based TPIA, is to build a DOCSIS compliant overlay on the existing proprietary cable plant. The concept is demonstrated in the following diagram.

Third Party Internet Access in Proprietary Systems

When cable operators first decided to provide Internet access on their cable systems, a few proprietary options were available to them from enterprising companies such as Terayon and Motorola. These systems were designed to provide data over the cable system as fast as possible, on a best effort basis, and priced to realize a return on investment for the operator in a reasonable time frame. They were not designed with the processing capacity or scalability to handle third party Internet access.

Proprietary devices achieved a reasonable degree of penetration in the early stage of data over cable deployment, and some eventually had some capability to provide third party access built into them as cable operators saw a market for it. Terayon, for example, built functionality into their proprietary CMTS that could be used to provide TPIA on a small scale. Simple next hop routing can be used in conjunction with modems pre-provisioned to be associated with a particular ISP to send packets to alternate service providers. Because of the simplicity of the routing capabilities of the CMTS, static routes are required to send traffic from the CMTS to the next hop device. This requires frequent intervention to keep correct routing tables as networks change and grow. The need for static routes is further complicated when multiple RF channels are connected to a CMTS, requiring blocks of IP addresses to be sub-netted between various outputs. In addition, routing hardware is required in close proximity (network-wise) to the CMTS, requiring additional expense by the MSO in routers to direct the IP traffic to the third party ISP.

While it is theoretically possible to provide TPIA over some cable systems using proprietary technology, it is not a practical solution. There is, first and foremost, the scalability problem. Internet service providers, particularly ISPs whose operating regions span multiple MSO systems, would need to adopt the same modem used by the MSO in each system; this would not offer the ISP much opportunity for service or price differentiation. Aside from the simple inability of some systems to allow third party access, such as Motorola, other technical issues that arise are:

• The proprietary systems were not designed with policy based routing in mind and, therefore, there is no effective way to assign a user and his/her SLA to a third party.
• Routing protocols for directing a specific item of IP traffic to a third party when dynamic IP address allocation (DHCP) is employed are not supported by the proprietary CMTSs, and, a third party access configuration relying on static routing would be difficult and expensive for MSOs to implement and maintain, and likely to be unstable.
• Billing and provisioning systems designed for use with proprietary systems have not been programmed to allow for third party Internet access.
• The types of features implemented in proprietary systems, such as next hop routing, are not scalable to large implementations of TPIA.
• Proprietary CMTS devices have been designed to quickly and easily provide data services over the cable system and do not have the processing capability to perform the large scale policy routing required for TPIA, even if the software could be programmed to do this (new DOCSIS based systems perform routing in hardware, making TPIA much simpler to implement at the CMTS).
• Due to the limited routing capability of proprietary CMTSs, providing TPIA via a router solution would require large numbers of routers close to the end user, or large-scale routers able to terminate many high-speed interfaces

With the observed trend of all major Canadian MSOs in moving to DOCSIS 1.0 devices with DOCSIS 1.1 hardware compliance, and the imminent upgrade to DOCSIS 1.1 compliance through software download, it seems unnecessary to spend the time, effort and money to develop complex, temporary solutions around proprietary equipment. New DOCSIS qualified CMTS devices have built-in routing and advanced billing support, easing the transition and reducing the costs of providing third party Internet access.

It is our view that regulating a DOCSIS solution to TPIA, as described in the section below, and allowing a reasonable time period for the MSOs to implement it, is the preferred option when compared with rolling out complex TPIA solutions with limited capability on existing proprietary systems. This is not to say that the MSOs should be simultaneously required to outfit their customers with DOCSIS compliant modems. Provided that it does not impede TPIA, the phasing out of their proprietary modems can take place on a schedule controlled by the MSOs' economic and competitive drivers.

Minimum Required Standard for MSO Cable Systems

In considering the question of imposing a minimum open standard to be met by all four major MSOs, we have formed the opinion that such an action could be taken by the Commission, and it would be generally beneficial to all concerned and not have a negative impact on the MSOs. They are already migrating toward DOCSIS 1.1 and are protecting their investment in their legacy facilities by implementing a DOCSIS overlay on their existing proprietary plant as described above. In our view, the minimum standard should be DOCSIS 1.1. Any future upgrades to their cable systems would have to be backward compatible to this level. Cable modems that support DOCSIS 1.0 would not be rendered obsolete by this requirement, since DOCSIS 1.1 is backward compatible.

After some period of time, it may be reasonable for the Commission to revisit the question of the minimum acceptable standard and set a new minimum standard, if DOCSIS 1.1 no longer meets the needs of MSOs and ISPs or it is no longer supported by the vendor community. It is expected that this step will not be necessary, as market forces acting on both the MSOs and ISPs will likely drive upgrades to higher minimum standards.

Imposition of rigid standards compliance can have the negative outcome that the industry's ability to innovate and advance the state-of-the-art is stifled. Looking at this matter from the MSOs' perspective, limiting the cable operators to strict DOCSIS compliance, without the ability to innovate, could hamper the advancement of the technology by preventing them from deploying new, desirable services made possible by alternative, possibly proprietary, systems offered in advance of the DOCSIS standards setting process. Permitting this kind of freedom, i.e., the ability to upgrade systems with new equipment, DOCSIS compliant, but with proprietary features, that can provide advanced services, seems desirable. A similar argument could be made to permit the third party ISPs to drive this technology advancement process as well but devising the rules under which this would be possible is complicated by the fact that the MSO is, after all, the owner of the network facilities, and by other commercial and competitive factors.

Formulating its ruling to maximize the ability to innovate would present the Commission with the rather difficult problem of creating a structure which is fair to both sides, one which, for example, protects the MSOs from unreasonable demands by the ISPs to incorporate ever-escalating and costly system upgrades, while, at the same time, assuring the ISPs unfettered access to cable customers with a competitive suite of services. The ability to incorporate proprietary features, over and above strict DOCSIS performance, could well prove to be beneficial to both parties. Standards are, after all, often set based on proprietary systems after they gain wide acceptance and commercial success.

Positive Trends and Tools for TPIA

Equipment Evolution

The most important barriers preventing third party ISPs from providing Internet access today are the non-compliance that MSOs have found with DOCSIS certified modems, and the continued use of closed, proprietary systems. These are important issues without simple solutions, since the Commission has already deemed it inappropriate for cable operators to specify the end user's cable modem³.

The trend in the Canadian cable industry is moving towards DOCSIS 1.1. In submissions to the CRTC, all of the major cable operators are either migrating to DOCSIS 1.1 from a proprietary system (when DOCSIS 1.1 hardware is readily available) or they are preparing to upgrade their DOCSIS 1.0 system when the new hardware is available (HSCO046, HSCO047, HSCO048, HSCO049). This industry-driven compliance with the DOCSIS specification will allow most MSOs to provide standardized third party Internet access, and eventually bring consumer choice to the purchase of cable modems. Regulations forcing adherence to a standard is required only if there are any cable operators who continue to operate closed systems. In general, as long as a cable operator can provide third party Internet access, they have met their regulatory obligations.

Cable modems that do not adhere strictly to the DOCSIS standards will continue to be a problem, until testing and certification standards are brought up to a level that prevents these non-standard modems from receiving certification. Test hardware exists with automatic testing scripts that can confirm MAC- and PHY-level standards compliance in a matter of hours. An important extension to this will be the higher level testing of network management information provided by the modems and the CMTS. As the industry continues to mature, products will be continually revised and cable engineers will become more knowledgeable in the design and test of these cable modem systems. It is reasonable to say that, as these systems become ubiquitous, the cable operators will have no choice but to allow any standards-compliant modem onto their network. To this end, they may have to design their systems with misbehaving modems in mind. Bandwidth usage and volume data may have to be tracked at the CMTS or POI router, rather than relying so heavily on the cable modem for billing and status information. Similarly, the CMTS will need to be do rate limiting, and be able to isolate and/or disconnect cable modems that are violating customer SLAs. CMTSs from manufacturers such as Cisco Systems have these capabilities today.

This trend is already being built on, as new OSS systems are designed to reduce truck rolls required for testing and troubleshooting cable modems. The ability to remotely characterize the power levels and manageability of a cable modem exists, and cable operators will find these systems invaluable as consumers begin to provide their own cable modems to attach to the cable network. Some of these companies and their products are highlighted in the next section.

As cable companies and ISPs begin to build service offerings around today's best-effort style cable services, CableLabs and the vendor community are already proposing new, faster methods of connecting subscribers to the Internet through existing cable systems. These faster protocols and infrastructure will have to be put into production as more and more subscribers compete for access and new, high bandwidth services, such as Voice over IP and Video-on-Demand, are offered over the same network. As business service is offered, SLAs will require the cable operators and ISPs to meet guaranteed bandwidth levels, also requiring QoS capabilities and broader bandwidths. It is important for any regulatory structure to take into account the future innovations in these networks, allowing for new, faster technologies but always maintaining a level of backward compatibility to allow ISPs, and even MSOs, to realize a return on the large investment in the current technologies.

OSS/BSS Software Developments

In reviewing the software being offered by players in this particular industry, we have noted a trend in ever increasing sophistication and capabilities of the software to not only provide for the more conventional customer assistance and management, but with the advent of DOCSIS 1.1, a high degree of sophistication is emerging. Some examples of software introductions at the 2001 NCTA cable convention include the following:

 

Manufacturer	Product	Description
1. BroadJump	Virtual Truck Installer	Automated installation and provisioning of broadband services, from qualifying the subscriber's computing environment to configuring the PC and the CPE to establishing and testing connectivity.
2. AP Engines	AP Interlink for High Speed Data Services	Automates the provisioning across multiple automated support systems and provides value-based customer billing.
3. Ceon Corp.	"It's On"	Service fulfillment solution, interfaces with front-end billing systems such as Siebel. It has a suite of event-based service fulfillment solutions that integrates software for order capture, order decomposition and billing systems.
4. Dorado Software	Red Cell Access Cable	Subscriber Management Service Provisioning & Activation Automated Assignment of IP Addresses Bandwidth Management SLA Management Flexible Report Writer Auto-Discovery and Inventory Management Fault and Performance Management of Network Assets
5. ChannelLogics	Network Access Management Software Suite	Monitors usage pattern, collects Network data and predicts individual usage of bandwidth and reports activities on network and individual subscribers.

This trend to ever-increasing sophistication in network management and control, and customer provisioning and qualification, appears to us to be heading in the direction of permitting cable operators to provide access to multiple, third party services providers, and to protect their own interest by being able to:

• Qualify modems remotely
• Test for technical capabilities
• Test for OSS capabilities that the cable company is relying on for control of its network, including the third party ISPs customers
• Police SLA agreements with third party ISPs and customers, and guard against wilful attempts to skirt payment for services provided
• Cause the CMTS to disconnect or isolate misbehaving modems
• Cause the CMTS to rate-limit modems that ignore the specified allowed bandwidth specifications

Development of these types of powerful tools for system operation should greatly ease the integration of third party ISPs onto HFC data networks.

Loosening of the Cable Modem Supply Chain

These trends to greater sophistication in network management software should also help the cable companies guard against the possible erosion of their ability to fully control their cable network through a trend to commoditization of the cable modem. Examples of this are:

• The incorporation of cable modems within computers. Computer-controlled-cable-modems, or CCCMs, are being developed by Comcast and Intel, and are expected to retail for $50.00 per unit). These are being designed to meet new CableLabs standards regarding computer controlled cable modems⁴.
• Some MSOs in the US, for example, AOL/ Time Warner, allow their customers to purchase their CMs at retail. There are guidelines as to makes and models and DOCSIS certification, but the tight MSO control of supplying the CM to the customer appears to be loosening. Their motivation for doing this is probably driven by a regulatory requirement to permit multiple TPIA on their cable systems or else lose the ability to offer AOL service, but the cable company is not the only supplier of the CMs. The cable company does, in most cases, recommend a make and model of cable modem that will work on their system, and provides no guarantees on the operation of other cable modems. See http://www.help.rr.com for the Road Runner specifications (Note that they vary from MSO to MSO).
• As well, Asian produced, low cost, clone-type cable modems will no doubt appear on the retail market for purchase by the public as more standardized, lower cost chip-sets are developed through large scale integration, for example, the new offerings by Texas Instruments and LSI Logic Corp.

These developments lead one to think that the current rigid compliance testing and tight control of which specific modem may be used to access the cable network may be further weakened over time, but with positive results. The developments described in the paragraph above would lead cable operators to place greater reliance on the ability of their management and operation systems software, an ability appearing in the kind of new offerings described in the previous section coming to market, to fully qualify a modem connected by a customer to the cable-based network. Powerful network management systems and operating support software, and more powerful CMTSs, may become as important as the DOCSIS standard itself in ensuring that cable modems purchased independently by customers and connected to the network, whether as a customer of the MSO or as a customer of a third party ISP, can be managed by the MSO and will enable the delivery of the requested services, and above all, will not be permitted to harm the MSO's network in any way.

The scenario presented above has, in our view, a very good chance of materializing, based on the historical behaviour of the development of other networking devices, for example, the dial-up modem. The importance of CableLabs' role in the testing (not standards setting and certification, which should still continue) of CMs may well diminish, because the variety of physical configurations and manufacturing entities of CMs might make it virtually impossible to control the testing process. Third party testing of CMs, on which we had earlier expressed the opinion as being viable, may well prove to be necessary and desirable in the environment of an open market supply chain for cable modems.

Conclusions and Answers to the Commission's Questions

This section is formatted to track the Commissions questions. The italicised text is taken from the Commissions scope of work document, followed in each case by our response.

1. Should the modems be DOCSIS compliant?

The Commission's decision applies to the four largest MSOs in Canada. Two of these are currently operating a DOCSIS 1.0 system. The others are using proprietary modems and have stated that they will wait until DOCSIS 1.1 is available before converting to DOCSIS and implementing competitive third party access. The conversion to DOCSIS will not be done on a flash cut basis so DOCSIS and proprietary modems can co-exist.

The technical trial for the third party access was carried out using DOCSIS modems. The third party access will be made possible by using policy based routing.

We would require advice on the technical feasibility and practicality of implementing competitive ISP access using proprietary modems.

Yes. MSOs should migrate their systems to a DOCSIS compliant system. The DOCSIS standard allows for open access cable systems and will allow consumer choice in service provider and cable modem hardware. Non-DOCSIS systems can remain in place as long they are TPIA capable, but should be upgraded to DOCSIS systems in the future.

Our research indicates that the technical and operational issues associated with establishing TPIA in proprietary systems are significant. Additionally, we offer the view that the situation is and will be complicated with the downturn in the economy and the delays associated with vendors achieving DOCSIS 1.1 compliancy, likely resulting in MSOs delaying or deferring DOCSIS deployments.

Some proprietary system vendors have implemented rudimentary capabilities in their CMTS products to provide third party Internet access, however, despite our enquiry, we have not located an implementation of this methodology.

Practical issues that impede the ability to provide TPIA efficiently are:

• The proprietary systems were not designed with policy based routing in mind and therefore there is no effective way to assign a user and his/her SLA to a third party.
• Routing protocols for directing a specific TP traffic to a third party when dynamic hosting is employed are not standardized and therefore difficult and expensive for MSOs to implement and likely to be unstable.
• Billing and provisioning systems designed for use with proprietary systems have not been programmed to allow for third party Internet access.
• The types of features implemented in proprietary systems, such as next hop routing, are not scalable to large implementations of TPIA.
• Proprietary CMTS devices have been designed to quickly and easily provide data services over the cable system, and do not have the processing capability to perform the large scale policy routing required for TPIA, even if the software could be programmed to do this (new DOCSIS based systems perform routing in hardware, making TPIA much simpler to implement at the CMTS).
• Due to the limited routing capability of proprietary CMTS's, providing TPIA via a router solution would require large numbers of routers close to the end user, or large-scale routers able to terminate many high-speed interfaces. While this is technically possible, the expense is not justified with the imminent move to DOCSIS compliance.

2. Which release of DOCSIS should be permitted (1.0 or 1.1)?

We are aware that there is some delay in certifying any DOCSIS 1.1 cable modems or CMTSs. As yet there is no confirmed date as to when any modems will be certified. We also are aware that many manufacturers are providing modems that are certified as DOCSIS 1.0 but are DOCSIS 1.1 "hardware compliant".

We would want to know:

i) whether it would be appropriate to require implementation of DOCSIS 1.0 compliant ( perhaps DOCSIS 1.1 upgradeable) CMTSs and use of DOCSIS 1.0 modems for the initial offering of ISP access for systems that have not upgraded to DOCSIS.

ii) whether there would be any problem with using DOCSIS 1.0 modems on a DOCSIS 1.1 systems.

Business success for all parties requires that a technology growth path be supported both by the regulator and by the industry as a whole. DOCSIS is a standard, which is evolving, and all certified levels should be allowed as economic and technical conditions allow. The current minimum level of DOCSIS device available on the market is DOCSIS 1.0 with 1.1 hardware compliance. It would be appropriate to regulate that TPIA systems meet this minimum level, with the option to upgrade in the future as long as this minimum requirement is still supported. Should DOCSIS 1.0 devices with DOCSIS 1.1 hardware compliance become obsolete, further proceedings can be held to decide on a new minimum standard for cable systems.

3. Whose certification for DOCSIS compliance should be required?

Is CableLabs certification for DOCSIS compliance required, if the specifications are public? As the MSOs are part owners of CableLabs, the view has been expressed that it might be preferable that another body do the certification.

CableLabs is the certifying arm for the industry and it has both the mandate and capabilities to correctly certify DOCSIS modems. CableLabs has a vested interest in the stability of Cable companies and is in an excellent position to protect the interests of their owners. As MSOs also have a vested interest in fostering a competitive environment for DOCSIS modems there is a natural driving force for them to certify a large number of modems.

It is our view that independent laboratory testing for DOCSIS compliance is workable. The process could be similar to that employed by other industries as, for example, certification of RF products for Industry Canada. Testing of a device is carried out by a recognized, independent individual or facility, which performs tests in accordance with published specifications that must be met. The test report is submitted to Industry Canada for review and issuance of a type approval certificate, if deemed compliant. Conceptually, a similar process could be used in the case of compliance testing for CMs and CMTSs.

Two prominent companies that specifically offer DOCSIS compliance testing as well as automated testing systems to facilitate such testing are Spirent and Agilent. Compliance testing for both DOCSIS 1.0 and 1.1 are offered for both CMs and CMTSs. There are other independent testing facilities, e.g., Tolly Group, (http://www.tolly.com), National Technical Systems, (http://www.ntsxxcal.com/home.html), and Interworking Labs Inc. (http://www.iwl.com), which offer DOCSIS compliance testing as well.

CableLabs themselves appear to be of this opinion that testing by outside labs is workable, since they have recently issued a Request for Proposals for outsourcing such testing though they did not accept any of the proposals. CableLabs has instead increased its testing capability for the new push to certify DOCSIS 1.1 systems. The concept of third party testing appears to be an option that CableLabs (MSOs) might accept.

Our discussions of this topic with MSOs revealed a general willingness, in the right circumstances, to accept outside testing by suitably qualified firms but there was also considerable reluctance as well.

There appears to be no fundamental reason not to permit third party testing provided that suitably qualified alternatives to CableLabs can be verified. CableLabs could continue as the certifying body by reviewing test reports prepared by independent test labs and issue compliance certificates, as is its current practice.

4. Once a modem has received the appropriate certification, is any further testing by the cable operator necessary?

Our order to the cable MSOs required that they allow connection of "compatible modems". The MSOs claim that in order for a particular modem to be "compatible" it must pass additional tests beyond the DOCSIS certification. We would like to know if this is in fact the case. We are aware that cable modem manufacturers submit to "second level" testing in order to meet the requirements of @home. The question is arises whether identified "second level" tests primarily relate to billing issues as opposed to technical operations?

Providing data over the HFC infrastructure is a very new technology, and the DOCSIS standards defining the technology are newer still, and evolving. At this time, based on the issues with software, hardware and new technologies, we believe that compatibility testing of basic operational functionality of cable modems by cable operators will be necessary for some time.

Testing of cable modems involves, among other things, analysis of the systems for management and billing and operations. MSOs emphasize their need to properly manage and query the cable modem for information. While certain information (RF power levels, for example) can only be obtained by querying the cable modem, other information (in particular bandwidth and usage statistics) can be monitored and maintained, and indeed policed, by most modern DOCSIS qualified CMTS systems and provided to a billing system or mediation system for subsequent processing. Cisco Systems, who have a significant portion of the DOCSIS 1.0 CMTS market, have implemented this functionality in their CMTS. Even if this was not the case, tools have existed for some time that could be implemented to monitor bandwidth and usage on IP networks in routers. As these tools are more integrated with cable systems, testing could be more specifically geared towards the functionality of the cable modem and the software to operate on the physical network. Bandwidth tracking for billing can be moved away from the customer premises and into the CMTS or router, where it is still distributed among many nodes, but allows for more certainty for the operator in the information provided.

We believe that at this stage of implementation of data over cable services and the imminent introduction of DOCSIS 1.1 capability, second level testing by the MSOs is necessary for the following reasons:

• Discovery of Unanticipated Problems
  It has been our experience that completely unforeseen interoperability problems frequently arise when a new element is introduced into a network. If a problem is anticipated, a procedure can usually be devised to test for its occurrence, but clearly, an unanticipated problem is hard to uncover in advance of equipment deployment into the network. Good engineering practice would dictate in-system tests prior to rolling out new network elements, particularly when the new elements are still in an early stage of development. Causes for this kind of behaviour are many and varied but often fall into one of the categories in the bulleted items below.

• Unique Hardware and Software Versions
  The inability of laboratory based tests to uncover problems that may surface in a particular cable system can be related to the extreme difficulty of reproducing a test bed sufficiently representative of a particular operator's cable plant and network in a lab environment. While test equipment may be able to exercise a particular modem in various respects, when similar functions are initiated by the particular versions of equipment and software existing in the field, the behaviour may be different than in the lab.

• Software Creep
  Manufacturers sometimes make minor modifications in software, and sometimes in hardware, between formal changes in version number. This is sometimes done to fix minor problems which have come to light after release of the product or in anticipation of a new feature in a subsequent release of the product. CableLabs' DOCSIS Certification Wave Guidelines state that vendors must "resubmit any certified product that is modified after certification"⁵, however minor software updates often do not go through this re-certification process, in particular when they are issued to fix bugs that are causing immediate problems with the MSO system. The cumulative effect of the introduction of several such modifications may result in undesirable behaviour in some circumstances, even though in the estimation of the manufacturer the changes did not warrant a re-certification.

• Manufacturing Tolerances
  Devices submitted for certification testing can exhibit tighter tolerances on certain parameters than units produced during subsequent, large scale production. While the specifications of units fabricated after certification testing would still be within the manufacturers tolerances, there can be sufficient difference to cause erratic behaviour.

• System Loading
  The behaviour of a contention based MAC layer is difficult to test in the lab for heavy network loading conditions. For this reason, purchasers of new or not fully hardened (mature) network elements will frequently include in the purchase contract a condition that the product is not accepted as compliant with the specification until it has been observed to operate properly in the field in a heavily loaded network for a predefined period of time.

• Increasing Complexity of DOCSIS Standards
  The configuration and setup of various customer/operator options, preference and operation parameters involve extremely large numbers of combinations and permutations of possible settings. It is almost impossible to test a piece of equipment sufficiently to arrive at complete confidence that it will operate properly in all MSOs' cable plants, operating different equipment and software loads, notwithstanding that they may be compliant to an open interface standard. The evolving nature of cable modem technology, with DOCSIS 1.1 implementations just around the corner, add even further complexity to the testing problem.

Eventually, it should be possible to greatly reduce second level testing by MSOs. but until the standards mature enough that this is possible, MSO second level testing should be permitted.

In the short term, it is our view that MSO qualification testing of third party cable modems should be permitted for a reasonable period to allow the technology to mature and to improve the scope of tests performed by CableLabs. As for the longer term, it is not possible for us now to predict when, if ever, second level testing by MSOs will become unnecessary. In the end it may still be necessary to permit the MSOs to conduct a minimum suite of tests in critical areas related to their specific network circumstances. In the future, the Commission could periodically review the state of technology relative to cable modems and decide whether continued MSO testing is necessary.

5. If any further testing is required what should these tests be and who should carry them out?

If additional testing is required, we would like to know what these tests should be. In the contributions provided in the CRTC CISC process, the MSOs have identified a number of additional tests that they submit are necessary.

We would also like to know if one set of tests would be required or whether the tests would in some way depend on the cable system. This also leads to the issue of who should do any supplementary testing. Would it be appropriate or necessary that each MSO carries out its own tests or could one central agency carry out the tests.

The contributions on this issue by the participants in High Speed Access subgroup can be found on the CRTC web site (www.crtc.gc.ca) at the CISC page under Ad Hoc groups -Cable High Speed Access).

We have observed that the position taken by cable operators, in general, is that they require the right to test and certify any product that has the potential to damage or degrade the pre-existing services on their systems. While there are many reasons cited for this, a strong position is held that DOCSIS compliance testing fails overall system testing when operating and business support systems are considered, i.e., factors unique to their own systems. Such systems are generally proprietary, complex and extensive and they are not DOCSIS certified but they do connect to DOCSIS compliant systems. We are informed that there is an initiative underway at CableLabs to create a standard format for billing data for DOCSIS compliant appliances, similar in function to the Bellcore AMA Format used in the telecom industry, and when this is completed, it should ease this area of testing.

As a matter of perspective, it is worth mentioning that in the USA where DOCSIS modems are widely deployed, DOCSIS compliant modems are openly available in a competitive consumer marketplace and are being used successfully in cable systems with limited further testing or regulation (however, most cable operators do recommend that consumers verify a make and model of modem with the MSO before making a purchase, to ensure complete compatibility, and each MSO has preferred models that they have tested on their system).

It is also worth noting that each MSO uses different cable network configurations and chooses different features provided in the DOCSIS certification, and that different vendors' CMTSs are qualified, rather than certified, allowing for different capabilities and product differentiation in the CMTS.

Based on these observations, we conclude that MSO testing of cable modems should continue for some time while the standard technologies mature. It is possible that some minimal amount of MSO testing will always be required due to unique cable plant parameters, equipment and software loads. The Commission could periodically review the maturity of the relevant technology to re-assess whether any level of testing continues to be necessary.

Glossary of Terms

BSS	Business Support Systems
CATV	Cable Television
CCCM	CPE Controlled Cable Modem
CL	CableLabs
CM	Cable Modem
CMCI	Cable Modem to Customer Premise Equipment Interface
CMTS	Cable Modem Termination System
CPE	Customer Premises Equipment
CRTC	Canadian Radio and Telecommunications Commission
DOCSIS	Data Over Cable Service Interface Specifications
HFC	Hybrid Fibre-Coaxial network
HSWG	High Speed Working Group
IP	Internet Protocol
ISP	Internet Service Provider
LSI	Large Scale Integration (chip manufacturing methodology)
MAC	Media Access Control (layer 2 of the OSI networking model)
MSO	Multiple System Operator
OEM	Original Equipment Manufacturer
OSS	Operational Support Systems
PC	Personal Computer
PHY	Physical layer (layer 1 of the OSI networking model)
POI	Point of Interconnection
QoS	Quality of Service
RF	Radio Frequency
SLA	Service Level Agreement
SNMP	Simple Network Management Protocol
TPIA	Third Party Internet Access
USB	Universal Serial Bus
VOD	Video on Demand
VoIP	Voice Over IP

Bibliography

Industry interviews

Agilent Technologies
AOL Canada
Cogeco Cable
Francois Menard
Passport On Line (former owners)
Rogers Cablesystems
Shaw Cablesystems
Terayon Communications
Videotron
Regional Cable

CRTC Documentation

CRTC Order 2000-789
CRTC Report HSRE006 "Unresolved Questions Concerning the Acceptance of Cable Modems," 19 June 2001
CRTC HSWG CISC Contributions HSCO001 - HSCO071

CableLabs Certification Documentation

All available from http://www.cablemodem.com:

DOCSIS 1.0, 1.1, and 2.0 Radio Frequency Interface Specifications
DOCSIS 1.0, 1.1, and 2.0 Operations System Support Interface Specifications
DOCSIS 1.1 Baseline Privacy Plus Interface Specification
DOCSIS 1.0 and 1.1 Baseline Privacy Interface Specifications
DOCSIS 1.1 Cable Modem to Customer Premises Equipment Interface Specification
DOCSIS 1.1 BPI Acceptance Test Plan
DOCSIS 1.1 CMCI Acceptance Test Plan
DOCSIS 1.1 OSS Acceptance Test Plan
DOCSIS 1.1 RFI Acceptance Test Plan
DOCSIS 1.0 Acceptance Test Plan
DOCSIS Certification Overview
DOCSIS Certification Wave Guidelines

Various World Wide Web Sites

http://www.cablemodem.com
http://www.terayon.com
http://www.cisco.com
http://www.ntsxxcal.com
http://www.iwl.com
http://www.tolly.com
http://www.agilent.com
http://www.spirent.com
http://www.cabledatacomnews.com
http://www.rr.com
http://www.caip.ca

Appendix A - Scope of Work

From: V. Lawetz
Telecommunications Branch
CRTC
Ph: 819-997-4578
eMail: victor.lawetz@crtc.gc.ca

Re: Consulting requirements: Cable Modems

Background:

The Commission currently has a process underway working out the details on how ISPs will be able to access Cable companies systems to provide high speed internet access. This process is the result of Order CRTC 2000-789 where the Commission established a policy for access by ISPs to the networks of the four major cable companies. A number of implementation issues were referred to the CRTC Interconnection Steering Committee (CISC) Cable High Speed Access Subgroup for resolution.

Issues

In its order the Commission directed the cable companies to permit the attachment of "compatible modems". One of the issues under consideration by the Cable High Speed Access Subgroup is the connection of different types of cable modems to the various systems.

There is disagreement amongst the parties about the conditions/criteria for allowing the different types of modems to be connected. The questions are:

a.) Should the modems be DOCSIS compliant?

The Commission's decision applies to the four largest MSOs in Canada. Two of these is currently operating a DOCSIS 1.0 system. The others are using proprietary modems and have stated that they will wait until DOCSIS 1.1 is available before converting to DOCSIS and implementing competitive third party access. The conversion to DOCSIS will not be done on a flash cut basis so DOCSIS and proprietary modems can co-exist.

The technical trial for the third party access was carried out using DOCSIS modems. The third party access will be made possible by using policy based routing.

We would require advice on the technical feasibility and practicality of implementing competitive ISP access using proprietary modems.

b) Which release of DOCSIS should be permitted (1.0 or 1.1)?

We are aware that there is some delay in certifying any DOCSIS 1.1 cable modems or CMTS's. As yet there is no confirmed date as to when any modems will be certified. We also are aware that many manufacturers are providing modems that are certified as DOCSIS 1.0 but are DOCSIS 1.1 "hardware compliant".

We would want to know:

i) whether it would be appropriate to require implementation of DOCSIS 1.0 compliant ( perhaps DOCSIS 1.1 upgradable) CMTSs and use of DOCSIS 1.0 modems for the initial offering of ISP access for systems that have not upgraded to DOCSIS.

ii) whether there would be any problem with using DOCSIS 1.0 modems on a DOCSIS 1.1 systems.

c) Whose certification for DOCSIS compliance should be required?

Is Cablelabs certification for DOCSIS compliance required, if the specifications are public? As the MSOs are part owners of Cablelabs, the view has been expressed that it might be preferable that another body do the certification.

d) Once a modem has received the appropriate certification, is any further testing by the cable operator necessary?

Our order to the cable MSOs required that they allow connection of "compatible modems". The MSOs claim that in order for a particular modem to be "compatible" it must pass additional tests beyond the DOCSIS certification. We would like to know if this is in fact the case. We are aware that cable modem manufacturers submit to "second level" testing in order to meet the requirements of @home. The question is arises whether identified "second level" tests primarily relate to billing issues as opposed to technical operations?

e) If any further testing is required what should these tests be and who should carry them out?

If additional testing is required, we would like to know what these tests should be. In the contributions provided in the CRTC CISC process, the MSOs have identified a number of additional tests that they submit are necessary.

We would also like to know if one set of tests would be required or whether the tests would in some way depend on the cable system. This also leads to the issue of who should do any supplementary testing. Would it be appropriate or necessary that each MSO carries out its own tests or could one central agency carry out the tests.

The contributions on this issue by the participants in High Speed Access subgroup can be found on the CRTC web site (www.crtc.gc.ca) at the CISC page under Ad Hoc groups -Cable High Speed Access).

Contract requirement

The contractor will submit a report addressing the above issues. The report should provide sufficient information for the commission staff to make a recommendation on what procedure should be followed to identify modems that are compatible with cable networks. The Commission must obtain right s to the report as it may be released to the public.

---

Footnotes

¹See Glossary of Terms for definition of acronyms used in this report. [return]

²Broadband Access Server - A large scale device that may act as a router or bridge designed specifically to aggregate large amounts of broadband data and forward it onto the backbone network, supporting various tunnelling, authentication, and QoS protocols. [return]

³CRTC, "Order CRTC 2000-789," CRTC: CISC [home page on-line]; available from http://www.crtc.gc.ca/eng/archive/2000/O2000-789.htm; Internet; accessed 10 October 2001. [return]

⁴CableLabs, "Cable Modem to Customer Premise Equipment Interface Specification," CableLabs [homepage online]; available from http://www.cablemodem.com/Specs/SP-CMCI-I06-010829.pdf; Internet; accessed 18 December 2001. [return]

⁵CableLabs, "DOCSIS^TM Certification Wave Requirements and Guidelines," CableLabs [homepage online]; available from http://www.cablemodem.com/DOCSISCertWaveGuidelines.pdf; Internet; accessed 18 December 2001. [return]

Date Modified: 2002-12-06