Methods To Detect And Resolve Rural Call Completion Problems In Real Time

Long Term Goals

The security and stability of the public telecommunications network has, as one of its foundations, the concept that any legitimate call from any device on any network will be capable of connecting to any other device on any network, irrespective of network operator, service provider, or user. One of the mandates of the Federal Communications Commission (FCC) is to ensure that all carriers provide such interconnect as would be reasonable to ensure such calls are capable of being completed. The all-IP network transition, from the circuit-switched public switched telephone network (PSTN), presents an additional opportunity to look for ways to enhance call completion.

As part of the PSTN Transition Program in the Georgetown site of the S2ERC, this project plans to conduct research into improving the security, reliability, accessibility, and utility of the nation’s telecommunications infrastructure.

Overview

The Federal Communications Commission (FCC) has found that problems of completing calls to rural areas, particularly areas served by rural incumbent local exchange carriers (ILECs) continue to be frequent and pervasive throughout rural America. The problems manifest themselves in a variety of ways, including lengthy periods of dead air on the calling party’s end after dialing a number, audible ringing tones on the calling party’s end when the called party’s telephone never rings at all, false busy signals, inaccurate intercept messages, the inability of one or both parties to hear the other when the call does go through, and calls simply not arriving at their destinations. The FCC has stated that the inability to complete calls reliably threatens public safety and contravenes the public interest. For example, the FCC has received examples of life-threatening call failures, including a situation where an on-call surgeon was unable to receive a call from a hospital for emergency surgery and a 911 call center was unable to do emergency call backs.

According to the FCC, there appear to be multiple factors that cause rural call completion problems. Rural associations (trade associations representing rural ILECs, or RLECs) posit that the call completion problems may arise from the manner in which originating providers set up the signaling and routing of their calls, and that many of these call routing and termination problems can be attributed to intermediate providers. The rural associations argue that least cost routing carriers offer terminating services at low rates, and that some least cost routing carriers may provide inferior service for a low rate.

The FCC has stated that one key reason for the increased problems in rural areas is that numerous providers in the call’s path often handle a call to a rural area. Given the particularly high rates long-distance providers incur to terminate long-distance calls to rural rate-of-return carriers, long-distance providers have additional incentives to reduce the cost of calls. For example, the disparity between interstate rates can be 5-6 cents per minute for rate-of-return areas and just over half a cent per minute for price cap areas. As a result, there is greater incentive for the long-distance provider to hand off the call to an intermediate provider that is offering to deliver it cheaply—and potentially less incentive to ensure that calls to rural areas are actually completed properly. The prevalence of these problems accords with providers’ incentives to engage in blocking or degrading traffic, or similar behavior, in an effort to minimize their intercarrier compensation payments.

Over the past several years, the FCC’s Enforcement Bureau has been investigating issues related to completion of calls to rural areas. In connection with those investigations, the FCC has entered into consent decrees with several carriers, including Verizon. As part of Verizon’s obligations under its consent decree, Verizon has agreed to engage in metrics-based investigations into potential rural call completion issues, and to fund academic research into methods to detect and resolve rural call completion issues in real time. This project will deliver on the required study.

The current state-of-the-art for rural call completion is to look at network data over a period of time, to try to identify anomalies in call completion or network performance, and then to take responsive action. Given the importance of these issues, there is a desire to try to remediate issues in as near-real time as possible.

Prior and current relevant work in the S2ERC includes the PSTN Transition Narrative project (identifying challenges and opportunities provider by the transition to the all-IP network), the Comprehendible Voice project (creating codecs to deliver superior comprehension rather than aural fidelity), the Database Transitions Study project (analyzing from a software engineering perspective issues surrounding the possible transition of the NPAC to a new vendor), and the Next Generation Caller Identification project (studying and creating technologies for providing secure caller identity assertions across service provider boundaries). In addition, there is a new project being discussed with S2ERC on the topic of combatting robocalling. Both the robocalling issue and the rural call completion issue have some similarities in the sense that both potentially involve data-driven measures as part of an overall solution, and both go to the integrity and availability of the PSTN. From a researcher’s perspective, there are a lot of synergies, parallels, and inverse relationships between the two projects, giving us economies of scale and personnel.

Intermediate Term Objectives

This project is in two phases. The first phase focuses on studying the models, metrics, and possible real-time remediation strategies carriers can implement to detect and resolve rural call completion problems. The second phase focuses on further data collection, refinements, and participation in Workshop 2.

Phase 1: Models, Metrics, and Remediation

Working with engineers and researchers from Verizon and possibly other carriers, the S2ERC will study examine existing collected data and current metrics to understand where gaps in data collection and processing lie. Our approach includes examining where important data is lost and if it can be recovered. We also expect to create and refine mappings between SS7, SIP, and our derived metrics to ensure the transition to SIP can only improve the collection of metrics and remediation of systemic call completion problems.

Today’s problem detection is either ad hoc, when a caller complains (unlikely), or post hoc, from running batch reports on call detail records (CDRs). Neither method is substantially real time. This draws us to a line of research where we can investigate whether the data is available, in the real-time signaling path, to collect meaningful metrics on call failures. For example, in SIP, if the remote end returns a 404 (Not Found) or 410 (Gone), there should be no surprise the call does not complete. There is no real user behind the telephone number. One thing to research is whether the rural carriers return proper and useful SIP response codes or SS7 ACM or RLC cause codes.

Likewise, proper handling by transit networks of rural carrier response codes also needs some study. For example, if the rural carrier responds to an IAM with cause code 25 (exchange routing error, usually because the rural carrier has an error in their translation tables), the transit network might respond with a cause code 3 (no route to destination) to the originating carrier. Depending on interpretation, this may be the correct behavior for the transit network (it does not know how to reach the destination) but to the originating carrier the code 25 indicates the number is bad, whereas code 3 may indicate a failure of the transit provider to offer connectivity to the rural carrier. One is an expected failure; the other is a failure to connect calls.

Determining the issues, failure modes, and data quality will require interviews with a number of participants including Verizon, trade associations, and other industry participants.

Moving to a real-time detection algorithm could help resolve a number of issues. For example, if a caller dials a number and immediately abandons the call, for example because they decided they did not want to place the call or they realized they misdialed, a real-time detection algorithm would know the call attempt and abandonment occurred in seconds, which would be a good indicator this was a user-directed abandonment. Conversely, if the caller dials a number and gets little or no call progress and after many seconds abandons the call, this could indicate a routing problem. As well, if the caller dials a number and the network does not report much call progress, yet the caller remains on the line for minutes, it is likely the call completed by the signaling was lost somewhere in transit. One feature of a few of the above-mentioned scenarios is a real-time detection algorithm does not strictly depend on the rural call signaling.

We expect Verizon will provide the bulk of the data necessary for these algorithms. We also expect that within a reasonable time period, working with Verizon’s engineers we could instrument some portion of their traffic to collect relevant metrics. We do not expect to deliver code certified to run as part of the active call path. As such, our definition of real-time is still some modest time period after the call succeeds or fails. Given today’s detection methodologies are on the order of days, we expect this to be a significant improvement.

Phase 2: Refinement and Workshop 2

Our expectation is that after analyzing the data already collected and prototyping new tools, we will, with the cooperation and collaboration of Verizon’s engineering staff, propose refinements to the tools.

We also plan on presenting our findings at Workshop 2.

Schedule of Major Steps

Phase 1:

  • Review existing FCC Notices, Orders, Decrees, and other documents relevant to the investigation
  • Review existing industry call completion metrics, including FCC data and ATIS metrics and methods
  • Review Workshop 1 proceedings
  • Review extant Verizon tools and data
  • Meet with Verizon to kick-off project, including an overview of newly available data, additional background information, and from Georgetown’s perspective, new ideas for data gathering and approaches for real-time resolution of rural call completion issues
  • Meet with other industry participants to get their input, guidance, and wisdom on rural call completion issues
  • Develop and deliver at least one system or method for detecting and resolving rural call completion issues

Phase 2:

  • If needed, refine systems and methods for detecting and resolving rural call completion issues
  • Produce paper on study results
  • Participate in Workshop 2

Dependencies:

Cooperation of Verizon and other carriers, including large MSO access providers, inter-exchange carriers, and rural local exchange carriers, for data and access to real-time facilities (or simulations thereof)

Guiding Principles:

We understand the desire for this research to result in systems or methods that could be put to practical use by carriers. Accordingly, in performing our research, we will be guided by the following considerations and will look for opportunities to propose systems or methods that:

  • Can detect potential call completion issues in real time;
  • Will work across a broad range of technologies, including SS7 and SIP signaling protocols and other industry standard or proprietary signaling protocols in common or wide-spread use through the communications industry;
  • Can be deployed efficiently and cost-effectively in an automated manner in carrier-grade networks and infrastructure;
  • Are simple to adopt and administer;
  • Minimize the incidence of false-positive events (indication of call completion issues where none in-fact are present);
  • Will work across different vendor platforms;
  • Can be based on or seamlessly integrated into existing carrier Operations Support Systems and environments, including existing industry standards for OSS and major switch vendor platforms; and
  • Have the potential to accurately identify the use of fraudulent or otherwise inappropriate call termination methods by carriers involved in the call completion process.