3GPP Release 99: The Birth of 3G
Release 99 is the specification that launched the 3G era. Published in 2000 and deployed commercially from 2001 onward, it defined UMTS — the first genuinely mobile broadband standard — replacing the voice-centric 2G world with a system built from the ground up to carry voice and data simultaneously.
Why is it called "Release 99"?
3GPP names releases by the year their feature freeze was targeted. This first release was frozen in 1999 — giving it the name Release 99 — even though commercial deployments began in 2001–2002. It is sometimes listed as Release 3 in certain 3GPP documents, reflecting its position as the third generation of mobile standards.
The Radio Access Network — UTRAN
UTRAN (UMTS Terrestrial Radio Access Network) sits between the mobile device and the core network. It replaced the 2G BTS / BSC pair with two new node types that offered much greater capacity and flexibility:
Node B is the physical antenna site — the 3G equivalent of a 2G BTS. It transmits and receives radio signals to/from the UE over the Uu interface using W-CDMA. Node B keeps its own intelligence to a minimum; all scheduling, handover, and routing decisions are made by the RNC above it.
The Radio Network Controller manages one or more Node Bs over the Iub interface. It handles handover decisions, uplink/downlink power control, and connects the radio access network to the core network via two separate Iu interfaces — one toward the circuit-switched (CS) domain, one toward the packet-switched (PS) domain.
The Radio Technology — W-CDMA
Rel-99 discarded GSM's TDMA (time-slicing) in favour of W-CDMA (Wideband Code Division Multiple Access). Rather than dividing the spectrum into time slots, W-CDMA lets every user transmit simultaneously across the same 5 MHz channel. Each user's signal is multiplied by a unique spreading code that expands it across the full 5 MHz bandwidth. At the receiver, the same code is applied in reverse — collapsing that user's signal back into a narrow band while reducing all other users' signals to low-level noise.
Key W-CDMA characteristics
- 5 MHz channel bandwidth — ten times wider than a GSM channel, allowing far higher data rates
- 3.84 Mcps chip rate — the speed at which spreading codes are clocked; determines how codes are generated and separated
- Variable spreading factor (SF 4–512) — a lower SF means more data bits per chip, giving higher throughput to that user at the cost of fewer simultaneous users
- Soft handover — a UE can be connected to two Node Bs at the same time, with both signals combined in the RNC, improving quality at cell edges
- Fast closed-loop power control at 1,500 Hz — transmit power adjusts 1,500 times per second on both UE and Node B to maintain consistent signal quality
- Peak data rates — 384 kbps outdoors (vehicular), up to 2 Mbps indoors (short range, pedestrian)
The Core Network — CS and PS Domains
One of Rel-99's most important architectural decisions was splitting the core network cleanly into two independent domains running on the same physical infrastructure:
Handles voice calls and SMS using a guaranteed, dedicated circuit for the entire duration of the call — the same model as 2G. Key nodes:
- MSC (Mobile Switching Centre) — routes calls within the mobile network and manages call state
- GMSC (Gateway MSC) — the interface between the mobile network and the public telephone network (PSTN), handling calls to/from landlines
Handles internet data by reusing the proven 2.5G GPRS nodes — no need to build a new packet core from scratch. Key nodes:
- SGSN (Serving GPRS Support Node) — tracks the UE's current location, handles mobility, and delivers packets between the access network and GGSN
- GGSN (Gateway GPRS Support Node) — the internet gateway; assigns IP addresses, applies QoS policies, and routes data to/from external networks
Separating the two domains allowed voice calls to remain rock-solid (CS reserves bandwidth exclusively) while internet data shared resources dynamically (PS uses bursts of capacity). Both ran over the same physical network infrastructure at the same time.
Quality of Service — Four Traffic Classes
Rel-99 introduced a formal four-tier QoS framework that let the network treat voice, video, web, and background data with completely different priorities and delay budgets — rather than putting everything into one undifferentiated pipe.
The strictest class. Both ends must stay in sync — any delay beyond ~150 ms becomes perceptible as awkward pauses. The network reserves bandwidth and delivers packets before anything else.
One-way continuous flow with a fixed, bounded delay budget. A receive-side buffer absorbs small jitter, so timing is less strict than Conversational — but it must remain consistent throughout the session.
Request-and-response pattern. The user expects a reply within seconds, but occasional variation in delay is tolerable. Error-free delivery matters more than tight timing — retransmission is acceptable.
Lowest priority — timing is irrelevant. Data just needs to arrive eventually and without errors. The network uses spare capacity for this class, making it very efficient for bulk transfers that the user isn't actively watching.
Security — AKA Protocol and KASUMI Cipher
GSM had a well-known security gap: only the network authenticated the device. The phone had no way to verify it was connected to a real base station — which allowed "IMSI catchers" (fake base stations) to intercept calls and track locations. Rel-99 fixed this at a fundamental level.
AKA — Authentication and Key Agreement
AKA replaces GSM's one-sided check with mutual authentication — both the network and the UE prove their identity to each other before any data flows:
- The network sends an Authentication Token (AUTN) that the UE can independently verify — confirming it is connected to a legitimate network, not a fake base station
- The UE responds with a signed Response (RES) — confirming to the network that the SIM card is genuine
- Both sides independently derive a matching set of session keys — CK (Cipher Key) for encryption and IK (Integrity Key) for integrity protection — without ever transmitting those keys over the air
KASUMI — Encryption and Integrity on the Air Interface
KASUMI (Japanese for "mist") is a 64-bit block cipher with a 128-bit key, derived from the MISTY1 algorithm. It powers two security functions that protect every transmission on the Uu interface:
- f8 — Confidentiality algorithm: encrypts both user data and signalling messages on the radio link, making eavesdropping computationally infeasible
- f9 — Integrity algorithm: generates a Message Authentication Code (MAC) over signalling messages, so any tampering or replaying of control messages is detected and rejected
Having both encryption and integrity protection on the radio interface was a substantial step beyond GSM, which offered encryption but no integrity protection on the air interface.
Voice Quality — AMR Codec
The AMR (Adaptive Multi-Rate) speech codec replaced GSM's older fixed-rate codecs with a system that continuously adapts to radio conditions. Instead of encoding voice at a single constant bit rate, AMR switches between eight codec modes ranging from 4.75 kbps to 12.2 kbps — selecting the best trade-off between quality and robustness for the current channel:
- Strong signal → higher bit rate (12.2 kbps) — more speech information per frame, more natural-sounding voice, closer to landline quality
- Weak or noisy channel → lower bit rate (4.75 kbps) — more redundancy and error correction bits replace speech bits, keeping the call connected even in poor coverage
- Mode switching happens automatically, multiple times per second, with no audible click or interruption to the user
The result is that 3G calls sound noticeably better than 2G calls across the entire coverage range — one of the most immediately tangible improvements users experienced when upgrading from GSM to UMTS phones.
Spectrum Flexibility — FDD and TDD
Rel-99 standardised two duplex modes under the same W-CDMA umbrella so that operators with different spectrum holdings could both deploy UMTS:
Uses two separate, paired frequency bands — one exclusively for uplink (UE to Node B) and one for downlink (Node B to UE). Both directions are active at all times simultaneously. The vast majority of global 3G deployments — Europe, Americas, most of Asia-Pacific — use FDD on the 2100 MHz band.
Uses a single frequency band, rapidly alternating between uplink and downlink time slots. More flexible for asymmetric traffic (e.g. more download than upload) and can operate in unpaired spectrum allocations. China's TD-SCDMA standard — used for its nationwide 3G rollout — is a variant of TDD-CDMA that Rel-4 later formalised.
Why Release 99 Still Matters
- Launched mobile broadband — for the first time mobile data was fast and cheap enough for real use: maps, email with attachments, basic video. The smartphone era would not have been possible on 2G data speeds.
- Established the CS/PS dual-domain architecture — this template was carried forward into every subsequent generation. Even 4G LTE's Evolved Packet Core reflects the design decisions made in Rel-99.
- Set the security baseline — mutual authentication (AKA) and air-interface integrity protection became the non-negotiable minimum for all future generations, with 4G and 5G building more layers on top.
- Created the upgrade path for HSPA — HSDPA (Rel-5) and HSUPA (Rel-6) are software and hardware enhancements layered onto the Rel-99 UTRAN. Operators did not need to rebuild their radio access networks to reach 14 Mbps speeds.
- Deployed globally at massive scale — by 2010, UMTS operated in over 170 countries. Its successor generations (HSPA, HSPA+) carried the majority of mobile data traffic until LTE became widespread around 2014–2015.