Voice vs Data Interconnection: Economics, Tech & Regulation

Historically, voice and data operated on separate networks (circuit-switched networks for voice and packet-switched networks for data). Technological convergence towards IP networks is blurring some of the physical boundaries, to the point that voice and data now share the same transport platform.

However, this technological convergence has not (at least as of today) dissolved the significant differences between the economic models, technical architectures and regulatory frameworks of these two interconnection services.

Differences in the economic model of voice interconnection and data interconnection

The most fundamental difference lies in how the traffic exchanged between data networks or voice networks is paid for:

• Voice (Calling Party Pays / Termination Monopoly): Voice interconnection has historically been governed by the “Calling Party Pays” (CPP) principle. Since a network has the exclusive ability to deliver a call to its own subscriber, a “termination monopoly” arises. This allows operators to charge other networks termination rates to complete calls. If left unregulated, these rates can be artificially high in order to generate revenue or harm competitors.
• Data (Bill-and-Keep / Peering): Internet (data) interconnection is mostly based on voluntary, commercial agreements known as peering. In 99.9% of peering cases, the model is settlement-free, or Bill-and-Keep. Here, networks exchange traffic free of charge on the basis of mutual benefit: the content provider wants to deliver data and the ISP wants its users to receive it. There is no natural “termination monopoly” for data, because content can be delivered over multiple routes or brought closer to the user through content delivery networks (CDNs).

Differences in technical architecture and network elements between voice networks and data networks

Convergence over IP networks

Voice transmission has evolved from legacy technologies (such as TDM, or time-division multiplexing) to IP-based architectures. This means that voice is digitised and split into data packets, travelling over the same fibre-optic cables, routers and switches as conventional internet traffic (email, web, video).

Since the transition of voice traffic to IP has not yet been fully completed, there are still infrastructure elements that act as bridges between modern networks and legacy networks. For example, Legacy Network Gateways (LNGs) are used to convert traditional voice traffic (TDM) into data format (SIP) so that it can travel over the new IP networks shared with data traffic, allowing old and new infrastructure to communicate during the transition.

IP networks were fundamentally designed to carry data without distinguishing between one type and another (net neutrality). However, voice traffic is a specialised application of this data transport and, in a sense, an exception to net neutrality.

Separation of signalling and media

In modern IP communications (which carry voice over data networks), there is a clear separation between signalling information (data used to control and set up the session, using protocols such as SIP) and media information (RTP, digitised audio and video). These flows travel along separate paths, unlike in traditional networks, where they tended to be more integrated.

Just like signalling (e.g. SIP) and media (e.g. RTP), “pure” data shares the same physical infrastructure (Ethernet cables, routers, switches, optical fibre). Everything is transmitted as IP packets over the same backbone network.

Traffic requirements and Quality of Service (QoS)

The main distinction is not whether they travel along separate paths, but how the network treats them:

• Media (Voice/Video): Requires low latency and low jitter. Routers and switches prioritise these packets through Quality of Service (QoS) mechanisms, ensuring they arrive quickly. In 5G networks, for example, target indicators are defined for ultra-reliable low-latency communications (URLLC), such as not exceeding 1 millisecond of latency, something that was not a priority in earlier data networks.
• Signalling (SIP): May also receive some priority to ensure that calls are set up and torn down quickly, but its volume is very low.
• “Pure” Data (Browsing, Files): Voice traffic now represents a tiny fraction of total network traffic, having been massively overtaken by data, and specifically by video, which accounted for approximately 82% of global IP traffic in 2021. Data is generally considered “best-effort” traffic and is not prioritised in the same way as voice. If the network becomes congested, data packets may be delayed or temporarily dropped to make way for the voice conversation.

Unlike the clear logical separation between the control plane (signalling) and the user/media plane (RTP for voice), “pure” data simply resides in the user plane and competes for overall bandwidth, subject to the QoS policies set by the network administrator.

Security and encryption

At the interconnection level, voice security (VoIP) measures and pure data security measures share many foundations (firewalls, VPNs, access control), but they differ in the specific protocols applied and the nature of the threats, especially with regard to the confidentiality of real-time conversations and call authentication.

Pure data security focuses on the integrity of the file or web session, using mature and widely adopted protocols such as:

• HTTPS (HTTP Secure) for encrypted web browsing, using Transport Layer Security (TLS).
• VPNs (Virtual Private Networks) to create secure tunnels across untrusted networks such as the Internet.
• File and Communications Encryption: Use of SSL/TLS for email, file transfers (SFTP/FTPS), etc.

Measures can also be applied to the content itself, such as antivirus and anti-malware tools that inspect the contents of data packets for threats (e.g. malicious attachments, phishing links).

Voice traffic, due to its real-time nature and its separation into a signalling plane and a media plane, requires specific measures at two levels (signalling and media), with protocols designed to ensure real-time confidentiality and caller identity authentication:

• SRTP (Secure Real-Time Transport Protocol): The specific protocol for encrypting the media stream (the audio), ensuring the privacy of the conversation.
• SIP-TLS (Session Initiation Protocol – Transport Layer Security): Used to encrypt SIP signalling traffic, protecting call setup data.
• STIR/SHAKEN: Emerging standards specifically designed to authenticate the caller’s identity and combat unwanted calls (voice spam and caller ID spoofing).

In modern corporate environments (such as MS Teams), voice over data networks requires specific security measures, such as signalling encryption (TLS) and media encryption (SRTP), requirements that were not native to traditional PBXs.

Interconnection and Peering: technical requirements to ensure connectivity

Traffic exchange between different networks (interconnection) takes place over peering and transit links that do not necessarily discriminate by type of content (voice or data), but instead connect autonomous systems to enable global connectivity:

• Internet Exchange Points (IXPs): Physical infrastructures where multiple networks (ISPs, CDNs, mobile operators) connect to exchange IP traffic, which may contain both VoIP voice calls and web browsing data.
• Common Protocols: The use of standard protocols such as SIP (Session Initiation Protocol) makes it possible to manage voice and multimedia sessions over existing data infrastructure, enabling services such as Microsoft Teams or Google Voice over the public internet.

Specific management elements

Although they share the same physical “pipes”, voice requires specific management within that shared infrastructure to ensure quality and security. This is done using Session Border Controllers (SBCs). These devices act as “borders” within the data network to secure voice traffic, manage signalling and ensure interoperability between different IP networks, making sure that voice (which is latency-sensitive) coexists properly with massive data traffic.

Data interconnection focuses on efficient routing through protocols such as BGP and the use of Internet Exchange Points (IXPs). Unlike voice, which requires an end-to-end circuit (virtual or physical) in real time, data delivery is optimised through caching (CDNs) within the ISP’s network or at nearby peering points, reducing the distance the data has to travel.

Regulatory frameworks for voice interconnection versus data interconnection

Because of their different economic models and the different founding philosophies behind them, regulators have treated these services differently:

Termination Monopoly vs. Voluntary Agreements

In voice networks, there has historically been a “termination monopoly“, since only one operator can deliver the call to the end subscriber. This has led to a regulated “Calling Party Pays” model, in which termination rates are charged to complete calls.

Because of the risk of abuse of the “termination monopoly”, regulators (such as the European Commission) have intervened for decades to set maximum voice termination rates (Eurotariffs) and ensure interoperability. Regulation guarantees that any user can call any other user (“any-to-any connectivity”).

By contrast, internet data interconnection is based on voluntary “peering” (traffic exchange) and “transit” agreements, which are usually settlement-free and aim for mutual benefit without the imposition of regulated termination rates.

Price Regulation

Wholesale voice termination prices have been heavily regulated and reduced over the years (for example, the TSLRIC+ model used in Australia) to prevent monopoly rents. In data networks, interconnection has thrived without price regulation, although some telecom operators (such as ETNO members) have recently proposed introducing mandatory network usage fees (“fair share”) for large data traffic generators, in an attempt to replicate the voice model.

There is strong harmonisation at European Union level. Mobile termination rates, for example, are capped by EU regulations, which has led to a significant fall in voice prices in recent years.

The IP interconnection market (transit and peering) has developed largely without regulatory intervention, based on commercial negotiations. Bodies such as BEREC have repeatedly concluded that this market works well on its own, thanks to competitive dynamics and the ability of networks to adapt (for example, during the COVID-19 pandemic).