Operation and Maintenance Center

In Figure 12.3 we showed the OMC as a network component with responsibility for the
physical transport links between the RNC and Node Bs, and legacy and 3G service
platforms. The RNC monitors the status of the transport links in the network—for
example, hardware or software failures. The network could be GPRS, EDGE, or UMTS
(UTRAN) or any corporate virtual private network implemented by the operator. As
with the RNC, this is a reasonably complex function. In theory, you should be able to
use one vendor’s OMC with another vendor’s RNC or Node B, but in practice there are
many vendor-specific variables in terms of implementation.

Summary
We described the major network components in a GPRS network (refer to Figures 12.5
and 12.6), including the SGSN and GSN (providing the gateway to other packet- or
circuit-switched networks). We also showed that Internet protocols are present but
not pervasive in existing networks, and that practical implementation, particularly
in 3GPP1 IMT2000DS/UTRAN networks, is very much based on ATM, both on the
copper access and radio access side.
E-GPRS uses higher-level modulation to increase the bit rate over existing GPRS
networks. E-GPRS also supports burst error profiles, which helps to make the radio
channel more adaptive and helps to reduce retransmission and retransmission delay.
GPRS and E-GPRS networks, however, do not, at time of writing, provide robust endto-
end performance guarantees and are unlikely to in the future, as this functionality is
not described in the standard.
ATM is increasingly pervasive as a hardware-based distributed switch solution for
managing a complex multiplex of time interdependent rich media data streams. The
multiplex carries on over the radio layer (which is effectively wireless ATM). 3GPP1
networks are not IP networks but, rather, ATM networks, supporting IP addressing
rather than IP-routed traffic streams. We return to this subject in Part IV of this book,
which is devoted to network software.
It is also difficult to see how GPRS can ever deliver sufficient dynamic range to support
highly burst offered traffic fired into the network from next-generation handsets.
3GPP1 determines a dynamic range excursion of 15 kbps to 960 kbps between two successive
10-ms frames. GPRS, as presently configured, is not able to support this.
Networks still using A-bis interfaces are also constrained on the copper access side.
ATM is needed on the IUB and IU interface for managing the incoming and outgoing
multimedia multiplex. This implies a significant upgrade to existing copper access
connectivity. Copper access quality and copper access bit rate flexibility are two necessary
preconditions for preserving rich media value.

Network bandwidth quality is dependent on network hardware quality. Software
routing is, generally speaking, insufficiently deterministic and insufficiently fast to
process bursty aggregated traffic as it moves into the network core. If IP protocols are
used, then substantial use of hardware coprocessing is required to deliver sufficient
network performance.
ATM (hardware-based circuit switching) is an alternative now being widely
deployed in 3GPP1 networks. It provides generally better measurement capabilities
than IP, which, in turn, makes it easier to implement quality-based billing. This suggests
that future network evolution may be more about optimizing ATM performance
over both the radio and network layer than optimizing IP performance.
On the one hand, we argued that future network value is very dependent on software
added value—our million lines of code in every handset. On the other hand, network
performance is still very dependent on network hardware, and radio performance is
still very dependent on radio hardware, which brings us to our next chapter. 297