In September, I had the privilege of planning and chairing the joint Society of Cable Telecommunications Engineers-Communications Technology voice over Internet protocol (VoIP) Symposium session on “Engineering VoIP with Capital and Operating Expenses in Mind.” During the planning process, we agreed to assemble a vendor panel where each member represented one of the PacketCable network elements, testing or operations support systems (OSSs) of a PacketCable-compliant system. One of the criteria for participation was that papers and presentations had to focus on expense engineering rather than technical implementation. Another criteria was that the presentations and papers could not promote individual products. Although the companies that employ the panelists have specific implementations to achieve capex and opex objectives, the panel deliverables were therefore required to be product agnostic. Presenting PacketCable in this context produced some creative, out-of-the-box thinking on the part of our vendor-panelists and essentially forced them to wear the customer’s hat. The panelists Six companies stepped to the plate to accept the challenge. Mike Clement, director of voice over cable solutions at Siemens, represented the call management server (CMS). Sonit Mahey, manager of system/application engineering at Cedar Point Communications, was the gateway person. Jay Strater, senior systems engineer at Motorola, spoke for the cable modem termination system (CMTS) and multimedia terminal adapter (MTA). Tim Spencer, president and COO at Sigma Systems, and William Yan, executive VP for sales and business development at Auspice Corp., handled OSS. James Pritchett, VP of engineering at Acterna, addressed testing. CMS The focus on economics directed panelists toward a business case approach. In his paper, Mike Clement highlighted how factors beyond the CMS must be considered in the overall business case for VoIP. He summarized the components of a detailed spreadsheet analysis that his company uses with operators to quantify the value of VoIP. “Siemens has identified nearly 75 cost drivers—some related to capital expenses, others to operating expenses and many interdependent on other factors,” he said. Per Clement, costs can be grouped into capex and opex equipment costs, service, repair, facilities and power, and staff training. Clement’s paper provided insight into how these factors are linked to the CMS. In it, he notes that CMS features, which normally fall on the revenue side of the return on investment (ROI) equation, must also be considered when calculating costs. As it points out, features drive network usage, which impacts the capital needed to meet network size and bandwidth requirements. As these factors increase, so do the operating expenses needed to keep them available. Furthermore, an operator must choose between a CMS with basic capabilities to complete calls or one that also includes a full range of services. Although the first alternative provides a lower cost initial capital expense, both capex and opex are needed later to add single-purpose servers with features such as CALEA and E911. Gateways Sonit Mahey’s contribution was an excellent illustration of the out-of-the-box thinking I mentioned earlier. Mahey’s company sells an integrated “PacketCable in a box” solution. For this meeting, however, he needed to discuss the pros and cons of both integrated and distributed gateway solutions. While still a bit biased toward the benefits of a single element management system, possible staff reductions by consolidation of functions at one location, and reduction on spares and licensing costs, he did point out cases where smaller systems might find advantage in a distributed solution. For example, when small media gateways with the capacity of a DS-3 or less are distributed over a dispersed geographical area, the cost of centralized CMS servers, recordkeeping servers and signaling gateways are shared by a large aggregate number of channels required for public switched telephone network (PSTN) handoffs at those locations. With line sizes below 20,000, a single link to a Signaling System 7 (SS-7) handoff point is adequate. After this point, additional links must be added, swinging the expense pendulum toward an integrated solution. CMTS and MTA Jay Strater presented some advice on how CMTS and MTA vendors could help operators manage capex and opex by considering tradeoffs during the design process. Low capital expense is achieved by designing CMTSs and MTAs with high efficiency but limited features. On the other hand, low opex comes by designing those same network elements with high VoIP quality of service (QoS) and traffic margins for reconfiguration when necessary. Node size and number of nodes per CMTS are among the critical parameters that need to be balanced. Jay pointed out that the resolution of conflicting objectives is optimized by an iterative design process, where the effects of the tradeoffs can be evaluated. In the future, he foresees more modularity in CMTS implementations, including decoupled downstream to upstream configurations. OSS Supporting and maintaining VoIP can have as much impact on expenses as the initial implementation. The importance of OSS was underscored by the fact that two vendors represented this area on the panel. Tim Spencer noted that VoIP brings many new challenges to an operator, including the need for integrated customer care and billing, order management, service architecture documentation, and network coordination. Self-provisioning and third party interfaces to an operator’s system make the task of delivering quality even more complex. Spencer pointed out that integrated service management produces significant staff savings through zero touch provisioning and will increase revenue by improved speed to market, reduction in customer churn and improved market differentiation. It will also extend the life of legacy investments by making use of existing infrastructure. The combination of these factors, according to Spencer, can result in as much as $10.80 of additional revenue per subscriber per year. William Yan reinforced this point, noting that time to market can be reduced from 18 months to 6 months, trouble tickets can be reduced by 15 percent, and ROI can be improved to reach payback in less than 6 months. He advocated a methodology that correlates alarms and trouble indicators across individual systems and network elements, including geographic grouping of trouble conditions to identify root causes. This methodology, said Yan, will create an efficient system where repetition and delays are eliminated and patterns can be studied to create proactive system maintenance. Testing Finally, James Pritchett spoke about the value of a comprehensive testing plan. I was struck by the fact he reiterated the same point that Ron Hranac emphasized in his Live Learning seminar a couple of months ago, namely that voice is not data. Voice has its own set of very stringent requirements for quality, and VoIP leaves no room for retransmission to compensate for lost information due to system faults. Again like Ron’s seminar, Pritchett’s talk noted that preparing for VoIP includes its own sweep testing and return path node certification and maintenance. In addition, Pritchett recommends testing at the subscriber’s home, segmentation and continuous monitoring, and optimization of work force efficiency. He pointed out that a well-thought-out testing plan can drastically reduce the three truck rolls typically required for marginal installation problems and that payback for a field instrument with a comprehensive test suite can be less than a year. There is a lot more detail in the symposium papers, which can only be summarized in the space I have available. If you are involved in system planning and did not get to attend, I highly recommend obtaining a copy of the papers or the DVD from SCTE for the full story. Justin J. Junkus is president of KnowledgeLink, an independent consulting firm. To discuss this topic further, you may email him at firstname.lastname@example.org.