TL;DR
This paper outlines the requirements for a universal end-to-end session identifier in IP-based multimedia communication networks. Such an identifier is crucial for robust troubleshooting, monitoring, and interoperability across diverse devices and protocols, addressing limitations of current call identifiers.
What Problem Does This Paper Address?
In today's IP-based communication landscape, services like SIP and H.323 use "call identifiers" to mark communication sessions. However, these existing identifiers fall short for several reasons:
- Protocol Incompatibility: SIP and H.323 use different syntaxes for their call identifiers, making it impossible to track a session end-to-end if both protocols are involved.
- Dynamic Changes: Devices like Back-to-Back User Agents (B2BUAs) often modify these identifiers, even within a single protocol, breaking the end-to-end link.
- Supplementary Services: Features like call transfers or session merging can further obscure the original session's identity.
This lack of a consistent, end-to-end identifier hinders effective troubleshooting, session tracking, and management across complex networks involving multiple devices and administrative domains.
Methods
The paper doesn't propose a specific identifier but rather defines the requirements for one. These requirements are derived from analyzing various use cases and constraints:
- Use Case Analysis: The authors examine scenarios like end-to-end identification, protocol interworking (e.g., SIP to H.323), traffic monitoring, tracking transferred sessions, and session logging.
- Requirement Derivation: Based on these use cases, specific requirements are formulated using RFC 2119 keywords (MUST, SHOULD, MAY). These requirements focus on the identifier's ability to:
- Be used by administrators and monitoring systems to track sessions.
- Correlate sessions that are transferred or joined.
- Pass through intermediaries unchanged.
- Avoid revealing sensitive information about users or devices.
- Work with devices that don't support the new identifier.
- Be unique in time and space.
- Be suitable for transmission in SIP and H.323.
- Have a constrained syntax for use in other protocols.
- Correlate sessions created via third-party call control (3PCC).
Datasets
Not discussed in the paper.
Key Insights and Ideas
- The "Call" vs. "Session" Nuance: The paper clarifies that "call" and "session" are overloaded terms. It defines a "communication session" for its purposes as the set of signaling messages related to a single invocation of a communication application between exactly two endpoints, from each endpoint's perspective.
- Intermediaries are the Problem: A significant challenge is how devices like B2BUAs and Session Border Controllers (SBCs) alter or strip existing identifiers, necessitating an identifier that can either pass through or be injected by these intermediaries.
- Privacy is Paramount: A critical requirement is that the session identifier must not reveal any personally identifiable information (PII) or network-specific details (IP addresses, hostnames, etc.). This prevents correlation of sessions with specific users or devices.
- Future-Proofing: The identifier needs to be designed with future protocols and interworking scenarios in mind, hence the emphasis on a constrained syntax and suitability for multiple transport mechanisms.
Key Results
This document is a requirements specification, not an implementation. Therefore, it doesn't present quantitative results. Its key "results" are the defined requirements themselves, which serve as a blueprint for future development of an end-to-end session identifier. These requirements aim to enable:
- Unified Tracking: A single identifier to track sessions across diverse network elements and protocols.
- Enhanced Diagnostics: Simplified troubleshooting and debugging.
- Improved Network Management: Better visibility into communication flows.
- Interoperability: Facilitating communication between systems using different protocols.
Limitations and Future Work
- Scope: The document focuses on requirements and does not specify the actual syntax or construction of the identifier. This is left for a subsequent document.
- Implementation Details: Specific security and privacy measures for implementing the identifier are outside the scope of this document.
- Legacy Systems: The paper acknowledges that legacy devices may not support the new identifier, and intermediaries will need to handle its injection.
- Future Work: The natural next step is the development of an actual identifier that meets these specified requirements, along with detailed security and implementation guidelines. Coordination with other standards bodies like the ITU-T is also highlighted as crucial for widespread adoption.
Summary updated