Network Slicing (5G)

Network slicing is a technique enabled by network function virtualisation and software-defined networking, under which the same physical 5G infrastructure, towers, spectrum, transport links and core network, is partitioned into multiple 'slices,' each behaving to its users as though it were an independent, dedicated network with its own performance guarantees. Because the underlying network functions run as software on shared computing infrastructure rather than on dedicated hardware, slices can be created, configured and torn down dynamically under software control, rather than requiring physical network build-out for each use case. Standardised 5G slice types illustrate why this matters commercially: enhanced Mobile Broadband (eMBB) slices prioritise high throughput for applications like video streaming; Ultra-Reliable Low-Latency Communication (URLLC) slices guarantee millisecond-level latency and very high reliability for applications like remote surgery, industrial automation or vehicle-to-everything communication; and massive Machine-Type Communication (mMTC) slices support very high device density at low data rates for Internet-of-Things sensor deployments. This capability, which requires a 5G standalone (5G SA) core rather than 5G deployed on a 4G core, is what distinguishes enterprise 5G, sold as a guaranteed performance envelope, from consumer 5G, sold as best-effort capacity. Network slicing sits in real tension with net neutrality principles, since a slice is, by design, a form of differential traffic treatment, prioritising some traffic over others. Regulators, including India's TRAI, address this through a 'specialised services' carve-out, which permits network optimisation for specific applications where the service is not usable as a substitute for general internet access and does not degrade general internet access, while requiring advance regulatory notification of new slices to police that boundary.

• 1 Network slicing partitions shared physical 5G infrastructure into multiple logically independent virtual networks with distinct performance guarantees.
• 2 Enabled by network function virtualisation and software-defined networking, allowing slices to be created/modified/retired under software control.
• 3 Three standard slice types: eMBB (high throughput), URLLC (ultra-low latency, high reliability), mMTC (high device density, low data rate).
• 4 Requires a 5G standalone (5G SA) core; distinguishes enterprise 5G (guaranteed performance) from consumer 5G (best-effort).
• 5 Creates tension with net neutrality, since slicing is inherently differential traffic treatment.
• 6 Reconciled through a 'specialised services' carve-out permitting optimisation for specific applications that don't substitute for or degrade general internet access.
• 7 TRAI's 2026 consultation paper proposed 21-day advance notification of new 5G slices, treating each as a separate tariff offering, to police the net-neutrality boundary.

A hospital's remote-surgery link could use a URLLC network slice guaranteeing millisecond latency, while an ordinary consumer video-streaming service uses an eMBB slice, both running on the same underlying 5G infrastructure but with entirely different performance guarantees.