Page 129 - From GMS to LTE

Universal Mobile Telecommunications System (UMTS) and High-Speed Packet Access (HSPA) 115

Such combined networks simplify the seamless roaming of users between GSM and
UMTS. This is still important today as UMTS (and LTE) networks in many countries
are still not as ubiquitous as GSM networks.
Seamless roaming from UMTS to GSM and vice versa requires dual‐mode mobile
devices that can seamlessly handover ongoing voice calls from UMTS to GSM if a user
leaves the UMTS coverage area. Similar mechanisms were implemented for data ses-
sions. However, owing to the lower speed of the GSM/GPRS network, the process for
data sessions is not seamless.
While UMTS networks can be used for voice telephony, the main goal of the new
radio access technology was the introduction of fast packet data services. When the first
networks started to operate in 2002, mobile network operators were finally able to offer
high‐speed Internet access for business and private customers. Release 99 networks
could achieve maximum downlink speeds of 384 kbit/s and 128 kbit/s in the uplink
direction. While this might seem to be slow from today’s perspective, it was an order of
magnitude faster than what could be achieved with GPRS networks at the time.
Accessing the Internet was almost as fast as over a 1 Mbit/s ADSL line – a standard
speed at the time. Since then, speeds in fixed and wireless networks have continued to
increase significantly and mobile network operators have upgraded their hardware and
software with features that are described in the following sections. The general network
design as shown in Figure 3.1, however, has remained the same.

3.1.2 3GPP Release 4: Enhancements for the Circuit‐Switched Core Network
A major enhancement for circuit‐switched voice and data services has been specified
with 3GPP Release 4. Up to and including Release 99, all circuit‐switched voice calls were
routed through the GSM and UMTS core network via E1 connections inside 64 kbit/s
timeslots. The most important enhancement of Release 4 was a new concept called the
Bearer‐Independent Core Network (BICN). Instead of using circuit‐switched 64 kbit/s
timeslots, traffic is now carried inside Internet Protocol (IP) packets. For this purpose, the
MSC has been split into an MSC‐Server (MSC‐S), which is responsible for Call Control
(CC) and Mobility Management (MM), and a Media Gateway (MGW), which is respon-
sible for handling the actual bearer (user traffic). The MGW is also responsible for the
transcoding of the user data for different transmission methods. This way it is possible, for
example, to receive voice calls via the GSM A‐interface via E‐1 64 kbit/s timeslots at the
MSC MGW, which will then convert the digital voice data stream onto a packet‐switched
IP connection toward another MGW in the network. The remote MGW will then again
convert the incoming user data packets to send it, for example, to a remote party via the
UMTS radio access network (Iu(cs) interface) or back to a circuit‐switched E‐1 timeslot if
a connection is established to the fixed‐line telephone network. Further details on the
classic and IP‐based circuit‐switching of voice calls can be found in Chapter 1.
The introduction of this new architecture was driven by the desire to combine the
circuit‐and packet‐switched core networks into a single converged network for all traffic.
As the amount of packet‐switched data continues to increase so does the need for invest-
ment in the packet‐switched core network. By using the packet‐switched core network
for voice traffic as well, network operators can reduce their costs. At the time of publica-
tion, most network operators have transitioned their circuit‐switched core networks to
Release 4 MSC‐Ss and MGWs. Figure 3.2 shows what this setup looks like in practice.