Designing The UTRAN Transmission Network

Once we have determined the number of required RNCs, based upon Iub
bandwidth requirements, we need to develop a homing plan that specifies
which base stations are to be controlled by which RNCs. This will define the
RNC borders. Analysis of the cells at or near those borders will then allow
us to estimate the amount of inter-RNC handover traffic we can expect, so
that we can determine the Iur connections required and the bandwidth
needed for those connections.
The exact amount of inter-RNC handover will depend on the RF environment
near the RNC borders. A reasonable approach, however, is to
assume that 50 percent of traffic in border cells is being served by two base
stations on different RNCs. This would be a conservative estimate that
would allow for additional inter-RNC soft handover involving cells that are
not defined on the border. Imagine, for example, a user near the top of a tall
building. That user might be served by a cell that is not on the border
between RNCs, but because of the user’s location, the user might also be
able to hear and be heard by a Node B on another RNC. Of course, the exact
soft handovers to be allowed in the network will be specified as datafill
within the RNCs. But, at the point in the design effort where Node B homing
and transmission network design are being performed, that datafill
may not yet be defined.
Given that the Iur acts in many ways as a conduit for Iub traffic from a
mobile to its controlling RNC, the basic assumptions for determining the
Iub bandwidth can be applied to determining the Iur bandwidth. For example,
if we assume that the Iub bandwidth needs to be approximately twice
the user throughput, then the Iur bandwidth should be close to twice the
user throughput for that portion of the traffic that is in inter-RNC soft
handover.
Now that we have established the Node B to RNC homing plan and we
know the Iub and Iur interface requirements, we need to design a transport
network to support all of the necessary connections between Node Bs and
RNCs, between RNCs and between RNCs and Service GPRS Support
Nodes (SGSNs) and Mobile Switching Centers (MSCs). Given that all of the
interfaces in question are ATM interfaces, we are effectively talking about
designing an ATM network.

In the example of Figure 12-9, we have three RNCs, each controlling a
number of Node Bs and with each RNC connected to each of the other two
RNCs. All three RNCs are connected back to a single SGSN and a single
MSC. In this example, all three RNCs are in different locations and all are
remote from the MSC and SGSN. This is a somewhat unrealistic situation,
but we use this example in order to show complexity and how that complexity
could be managed. Figure 12-9 shows the logical connections
between the various nodes.To implement each of those interfaces individually,
however, would be impractical. Rather, one would like to implement a
cost effective transport arrangement that will support each of the logical
interfaces. One way to do this could be through the use of a ring arrangement
as also shown in Figure 12-9. Basically, each of the locations in question
would become nodes on the ring, which might be an OC-12, or perhaps
an OC-48 ring or even have a higher capacity depending on demand.
In many cases, however, the distances between nodes could mean that
the cost of such a ring could be prohibitive. In that case, one might want to
employ one of the configurations shown in Figure 12-10. In the first case,we
deploy a separate ATM switching layer that takes care of switching of the
various ATM paths between the various nodes. By deploying such a layer,
we can reduce the overall transmission cost. Of course, there is a capital
cost that must be paid, plus the operational cost of deploying new equipment.
In the second configuration of Figure 12-10, we use one of the RNCs
as an ATM switch. Back at the MSC site, we may have the possibility to use
an SGSN or an RNC at that site as an ATM switch. This option is possible
for some equipment vendors because an RNC is fundamentally an ATM
switch with additional UMTS-specific functionality. It is not uncommon to
find that the total switching capacity of an RNC is several gigabits per second,
while the Iub interface capacity may be limited to perhaps 200 Mbps.
Thus, we are likely to find that the RNC can switch more ATM traffic than
would be required of it as a pure RNC.We can take advantage of this extra
switching capacity and reduce overall transmission cost without having to
deploy a separate ATM switching network.
The design and cost of the UMTS Terrestrial Radio Acess Network
(UTRAN) transmission network is interwoven with the placement of the
RNCs. There may be multiple options for placement of RNCs. We may
choose to place all RNCs at the MSC location, all remotely, or some mix of
remote and local RNCs. The placement of the RNCs will be related to the
capacity of an RNC, the cost of the RNC, the availability of suitable locations,
and the cost of transmission. The final solution must aim for a network
topology that strikes a balance between capital cost, operational cost,
and network reliability.