Interface Design and Transmission Network Considerations

In general the determination of the amount of bandwidth required for a
given interface is a relatively straightforward process. For example if we
expect a given RNC to support a given number of Erlangs, then we can easily
determine the required bandwidth on the Iu-CS interface. Similarly if
we size some RNCs to support a given amount of data traffic in the busyhour,
then we can determine the bandwidth required for Iu-PS interface,
and so on.Thus, once we have used traffic forecasts and dimensioning rules
for the access network elements, we can determine the bandwidth requirements
from the access network to the core network.

For example if a given RNC is expected to carry 2,000 Erlangs of voice
traffic in the busy hour and if we dimension the Iu-CS interface at a 0.1 percent
grade of service, then Erlang B tables tell us that we need approximately
2,100 circuits. Assuming that traffic between the RNC and the MSC
is carried at 16 Kbps, we need 2100/4 DS0s, which equates to 525 DS0s, or
approximately 22 T1s. This bandwidth is carried over ATM, so we must add
approximately 201 additional overhead for ATM.
Similarly, if a given RNC is expected to carry 50 Mbps of user data, then
we can directly determine the Iu-PS bandwidth requirements. It will typically
be about 120 percent to 130 percent of the user data bandwidth to
enable for GTP overhead. Thus, for 50 Mbps of user data, we would need a
bandwidth of 65 Mbps on the Iu-PS interface. Again, ATM overhead must
be added.

Dimensioning of interfaces between RNCs (or between BSCs) will
depend on the specifics of the radio network. RF design input regarding the
amount of soft handover traffic is critical. For example, if RF designers estimate
that 20 percent of all voice calls will involve inter-RNC soft handover,
then we can use that information to determine the bandwidth requirement.
For example, we can assume that 20 percent of the voice traffic will be carried
across a given Iur interface.

Overall, the dimensioning of bandwidth requirements for individual
interfaces is not overly complicated provided that we have determined the
number of network elements and have established detailed traffic demand
estimates. The next step, however, is the design of a transport network that
supports those bandwidth requirements in an efficient but reliable manner.
That design effort can involve more complex issues and involves a greater
degree of network design expertise. Consider, for example, a scenario such
as is shown in Figure 10-1. In this example there is a large local market and
a remote medium-sized market. It has been determined that one MSC and
three SGSNs should be placed in the large market. These are connected to
four co-located RNCs that serve the local market. In addition, there are two
RNCs placed in the remote market. Thus, we need Iu-CS and Iu-PS connections
from the remote market to the local market.We may also need one
or more Iur interfaces between the local market and the remote market,
particularly if the RF coverage of an RNC in one market borders the RF
coverage of an RNC in another market, as might be case along a highway
between the two cities.

In addition, in North America at least, there will need to be connections
from the MSC in the local market to the Public Switched Telephone Network
(PSTN) in the remote market for support of PSTN calls to or from subscribers
whose dialable numbers belong in the remote market. Imagine for example a subscriber in the remote market who makes a local call. That call
is first carried to the MSC and then must be carried back to the PSTN in
the remote market. This can be done through direct trunks to that PSTN
carrier as shown in Figure 10-1, or the calls can be handed over to a long
distance carrier.

Finally, of course, there needs to be hundreds of Iub interfaces from the
RNCs to the base stations in each city. All of these interfaces and the associated
bandwidth requirements need to be supported by an integrated
transmission design that provides the necessary bandwidth and reliable
transport.
In most cases, the MSC will be placed on a fiber ring. The total capacity
of the ring will depend on the total transmission in and out of the MSC site,
but it will be divided into a number of discreet capacities such as a number
of DS3s or OC-3s. Typically, the ring will have a number of nodes, including
hubbing nodes that belong to the ring provider. The individual links from
the base stations in the local market will be connected to such points on the
ring where they will be multiplexed onto DS3s or OC-3s for transport to the
RNCs at the MSC site.

At the remote market, there will also be a significant number of Iub
interfaces from base stations. Depending on the number of such interfaces,
the availability of transport facilities, and the cost of those facilities, the
remote RNC site might also be placed on a ring. In fact, if the distance is not
too great, the remote RNC location might be a node on the same ring as the
MSC site. In many cases, however, the distance between the cities may be
too great to justify the cost of extending the ring to the remote city. In any
event, the traffic from the remote RNC location to the MSC location will
need to be protected. This will generally mean that there are diverse transmission
facilities between the remote city and the MSC location. These
diverse facilities must be sized to support the Iu-CS, Iu-PS, Iur, and PSTN
connectivity requirements. This diversity will involve extra cost. That extra
cost, however, will generally be justifiable. After all, the traffic demand in
the remote city will be significant or it would not have made sense to dedicate
RNCs in that remote city.

Often, the availability of transport facilities is a major factor in the
timely deployment of a network. In the United States, most transport uses
transmission facilities leased from local- and long-distance carriers.
Depending on the carriers, it might take six months before a ring can be
installed at an MSC location. Moreover, it may be necessary to wait until
the ring is installed before ordering individual circuits on that ring. Consequently,
the earlier the transmission network requirements can be established
the better.