Professional Documents
Culture Documents
Chapter 3
OBJECTIVES:
Upon completion of this chapter the student will be able to:
Intentionally Blank
Topic Page
NETWORK TOPOLOGY....................................................................115
NETWORK ARCHITECTURES ................................................................................. 116
LARGE REGIONAL OR NATIONAL NETWORKS.................................................... 118
BENEFITS OF USING TRANSIT EXCHANGES COMPARED TO A MESHED
NETWORK WITH DIRECT ROUTES ........................................................................ 120
TRAFFIC ROUTING...........................................................................132
PLMN TO PSTN......................................................................................................... 132
PSTN TO PLMN......................................................................................................... 134
GSN ROUTING METHODS....................................................................................... 136
IP ROUTING .............................................................................................................. 136
Intentionally Blank
NETWORK TOPOLOGY
When planning the network architecture, it is important to take the
future expansion of the network into consideration, so that a stable
and preplanned development of the network architecture can be
achieved. Changing the network architecture is costly and the time
required for migration into a new architecture is considerable. It
requires a large number of intermediate reconfigurations and re-
dimensioning, if the network is large. The architecture planning
should, therefore, have a time horizon of approximately three to
five years and be co-ordinated with the general goals of the PLMN
operator’s business plan.
• Number of MSCs
• Transmission costs
• Traffic distribution
• Traffic volume
• PSTN tariffs
• Availability requirements
The network topology plan should present the nominal design of the
network, i.e. the functionality to be located at each site chosen, the
structure of the network, the approximate number of GSNs to be
located per site chosen and the type of connectivity between sites. This
design is termed “nominal”, as further design and dimensioning must
be carried out, and this would normally involve some alterations and
refinement to the Network Topology.
NETWORK ARCHITECTURES
The basic examples below show possible traffic network
configuration principles for small (1 - 4 MSCs), medium sized
(5 - 10 MSCs), and large (>10 MSCs) regional or national
networks.
SMALL NETWORK
The diagram below represents an example of a small network
with three GMSCs. A simple meshed network with trunks
connects the GMSCs. Each MSC is connected to the PSTN.
G G
INT INT
TGMSC TGMSC
G G G
P TG TG P
other other
G G G G
P P P P
T T
P P P
• Increased flexibility
• Easily expandable network
• Platform for functional development
• Lower handling costs
• Improved signaling network
Increased flexibility
In a situation where it is difficult to forecast subscriber growth,
subscriber behavior, transmission costs and switching
equipment, it is important to build a network that is very
flexible.
TRANSIT TRANSIT MS C MS C
MSC MSC MS C MS C
INTRODUCTION
The general topology for any GPRS network can be designed, and later
dimensioned, using individual components including GSNs, Routers,
FR Multiplexers, TDM Cross-connects, Ethernet Switches, Servers for
Management and GPRS Support Services (e.g. RADIUS, DNS, DHCP,
NTP), PoS Links, ATM Links, FR over E1 Links, etc. The combination
of components possible at any site in the network could vary greatly.
SITE TYPES
The site types that make up a GPRS Network are described below:
Services LAN combines application servers and GPRS services like NTP,
DNS, Billing Gateway, RADIUS, and Network Management.
Core Site with one or more GGSNs (or CGSNs) with normal IP Routers or
one Carrier-Class Router as core element. SGSNs may be co-located. The site
can contain ATM-Switches for Gn and Gb connectivity.
Core Site with one or more GGSNs (or CGSNs) with normal IP Routers or
one Carrier-Class Router as core element. SGSNs may be co-located. The site
can contain ATM-Switches for Gn and Gb connectivity.
Concentrator site for BSC/PCU and corporate or ISP access with no core
functionality and no ATM switching capabilities. Concentrator sites are
connected to secondary or primary sites
For SGSN and GGSN sites, it is normal (for Ericsson GSM equipment)
that these are also BSC/PCU sites, i.e. Ericsson BSCs originally located
there, PCUs added, as well as GSNs.
The table below gives a set of criteria on which to select the Site Type
required at any location. Detailed guidance is given the section on
Topology Design Process.
Concentrator No No No No No No Yes
Site
Inputs
The following inputs are required for the planning of the network topology:
Outputs
• Number of Regions,
• Number of Sites,
• Type of Site (Primary, Secondary, Concentrator),
• Hierarchy
• Connectivity between sites
10.Consider the physical topology of other related networks – GSM nodes, SDH /
Leased Lines, ATM, FR or other data networks, Interconnect point to other
PLMNs. This helps to identify node co-location possibilities and to recommend
the location of GSNs.
CENTRALIZED TOPOLOGY
In the Centralized Topology, the FR network will constitute the
major part of the network, with the IP backbone for Gn
interfaces internal to the central site (probably Ethernet based).
A number of frame relay switches or multiplexers will be placed
in each region and connected to two or more centralized FR
hubs, co-located with the GSNs and feeding E1s into them. The
interfaces between regional FR device and the hubs can be E3 or
STM1 (ATM).
The FR network from the regions to the central hubs can use a
double star (or dual-homed) architecture. It should include
resilience by having two diverse links to each of the central hubs
(see figure below). These links can be protected in the transport
(ATM or SDH) level.
GPRS Network
FR
Services LAN
FR network to
FR External
other BSC/PCU CGSN Networks
sites
e.g. ISP,
Corporate
FR
R
Gb FR on E1s CGSN
The FR
FR
backbone
network R
Indirect Gi
FR
CGSN DirectGi Various WAN
links as
required.
Gb FR on E1s R Gi Forwarder for the
Ethernet indirect Gi
DISTRIBUTED TOPOLOGY
The FR interface will only be used in the regions (to the SGSN
in the same region). The IP backbone will form the major part of
the network and will be described below in Gn section.
External Networks
e.g. ISP, Corporate
Backbone Network
CGSN
CGSN CGSN CGSN
MIXED TOPOLOGY
The FR interface will only be used in the regions (to the nearest
SGSN, which may be in the same region or another region).
For all topologies, if BSC / PCUs are co-located with the SGSN
and sufficient E1 interfaces are available on the SGSN to handle
the Gb capacity (without any reduction due to multiplexing)
then FR multiplexers may not be justified. If the number of E1s
is small, even if they are not co-located then the cost of the
equipment vs. the cost of extra transmission saved should be
considered carefully.
Secondary Site
Secondary Site
Secondary Site
CGSN
CGSN CGSN
GPRS Network
Services LAN
GPRS Network
Services LAN CS CS CS CS
BSC / PCU
AR AR
Ethernet
External Networks External Networks
FR Gb e.g. ISP, Corporate e.g. ISP, Corporate
BSC / PCU
TRAFFIC ROUTING
We will consider two MSCs, which we assume to be located in
two different cities A and B, see below.
PLMN TO PSTN
Often, the PSTN operator requires the PLMN operator to
implement a PSTN network routing analysis in the MSCs and to
direct the traffic to specific PSTN exchanges within a city area,
as previously mentioned.
Alternative 1.
Flat rate is applied, which means that the cost of carrying traffic
in the PSTN is distance independent. The traffic should,
therefore, be dropped to the closest possible PSTN transit
exchange. This is called near-end drop.
Alternative 2.
When traffic has reached the far end MSC, it is routed to the
PSTN. This routing principle is called far-end drop.
TE TE
PSTN TO PLMN
The most frequent way of routing mobile terminated traffic from
PSTN subscribers is illustrated in the figure below. In this case,
the PSTN operator drops the traffic from PSTN subscribers at
the closest possible GMSC. One reason for this routing principle
is that the MS's MSC/VLR is unknown to the PSTN exchange.
When the call has been handed over to the PLMN, the most
common PLMN/PSTN operator-to-operator agreement is that
the call should be kept within the PLMN, if the call is
terminated in the PLMN. If the call is redirected to a PSTN
subscriber using a PLMN redirection service, a call transfer
back to the PSTN is allowed.
RIP
OSPF
BGP
IP ROUTING
One of the basic functions of IP is its ability to form connections
between different physical networks. This is due to the
flexibility of IP to use almost any physical network below it.
Routing is the act of moving information across an Internet from
the source to a destination. Routers interconnect the various
network segments making up the Internet. A router receives an
IP packet on one of its interfaces, and forwards the packet out
on another of its interfaces (or possibly more than one if the
packet is a multicast packet), in accordance with the contents of
the IP header. Routing from the source IP address to the
destination IP address is performed by the IP forwarder. The IP
forwarder performs this function by extracting the destination IP
address from the IP packet and comparing it to entries in a list of
IP addresses, each with an affixed path. The path is primarily an
identification of the correct router’s physical port to send the IP
packet. The list of IP addresses used by the IP forwarder is
contained in the route table (RT).
IP Routing
Source Intermediate Destination
router router router
Internet Internet
Link Link
Physical Physical
Static routing
Dynamic routing
Routing Protocols
Dynamic Routing
Interior Exterior
Internet
BGP
area-2
OSPF
area-0
area-1
RIP area-3
OSPF
area-0
Each router saves the most recent table it receives from each
neighbour and uses the information to calculate its own distance
vector table.
The distance vector table (the routing table) for the router is then
created by taking the lowest cost calculated for each destination.
RIP RIP
UDP UDP
Protocol 17 Protocol 17
IP IP
RIP data grams are normally broadcast onto LANs using the LAN
MAC all-stations broadcast address and the IP network or subnetwork
broadcast address. They are specifically addressed on point-to-point and
multi-access non-broadcast networks, using the destination router IP
address. Routers normally run RIP in active mode; that is, advertising
their own distance vector tables and updating them based on
advertisements from neighbours. Endnodes, if they run RIP, normally
operate in passive (or silent) mode; that is, updating their distance
vector tables on the basis of advertisements from neighbours, but not in
turn advertising them. RIP specifies two packet types: request and
response. A request packet is sent by routers to ask neighbours to send
part of their distance vector table (if the packet contains destinations),
or all their table (if no destinations have been specified). A response
packet is sent by routers to advertise their distance vector table in the
following circumstances:
• Every 30 seconds
• In response to a request packet
• When distance vector tables change (if triggered updates
are supported)
route table
route table
30 180 seconds
seconds (maximum)
(minimum) >180, set inactive
120 seconds
(delay) purge from RT
Active and passive systems listen for all response packets and update
their distance vector tables accordingly. A route to a destination,
computed from a neighbours distance vector table, is kept until an
alternate is found with lower cost, or it is not re-advertised in six
consecutive RIP responses. In this case the route is timed out and
deleted. When RIP is used with IP, the address family identifier is 2 and
the address fields are 4 octets. To reduce problems of counting to
infinity the maximum metric is 16 (unreachable) and directly connected
networks are defined as having a metric of one. RIP makes no provision
for passing subnet masks with its distance vector tables. A router
receiving a RIP response must already have subnet mask information to
allow it to interpret the network identifier and host identifier portions of
the IP address correctly. In the absence of subnet mask information a
router will interpret routes as best as it can. If it knows an IP network
has a specific subnet mask, it will interpret all other route information
for that network on the basis of that single mask. If it receives a packet
with bits set in the field that it regards as the host field, it will interpret
it as a route to a host with a mask of 255.255.255.255. The above
makes it impossible for RIP to be used in an internetwork with variable
length subnet masks.
OSPF OSPF
Protocol 9 Protocol 9
IP IP
LSD
LSD LSD
identical
LSD
BGP BGP
TCP TCP
Protocol 6 Protocol 6
IP IP
Traffic Theory
A = y. s Erlang
Grade of Service
Busy hour
PDP context
Traffic measurement
(100 000 * 20) + (87 500 * 25) + (62 500 * 20) = 5,437,500 mE
The IPSec column in the tables below are based on the outer IPSec Header
for Transport mode, if Tunneling is used an additional 20 bytes have to be
added.
For FR the Ericsson GSN product supports DL-Core over PVCs and has
the ability to run FR either as DTE (towards a FR network) or DCE for
point-to-point connections in the case where there is no FR network.
Table 3.5
The Gn Interface Protocol Stack
GTP (GPRS Tunneling Protocol) tunnels user data and signalling between
GSNs in the GPRS backbone network. GTP encapsulates all point-to-
point (PPP) Packet Data Protocol Packet Data Units (PDUs). GTP
provides mechanisms for flow control between GSNs if required.
UDP (User Datagram Protocol) carries PDUs for protocols that do not
need a reliable link. IP (Internet Protocol) is the GPRS backbone protocol
used for routing user data and signalling.
Table 3.6
The Gi interface is the connection from the GGSN towards the external
networks e.g. ISP or corporate LANs. The choice of interface is with the
operator but ATM or Ethernet are recommended.
Table 3.7
The Gp traffic comes from each SGSN and GGSN. It is carried to the BGs
(Border Gateways). Traffic over the Gp is in two parts: traffic generated
by other PLMN customers roaming in your network and traffic generated
by your customers roaming in other PLMNs. These are calculated
separately and summed.
Capacity calculation from each SGSN and GGSN to BGs, the overhead is
similar to the Gn case above. The network capacity is dimensioned for full
redundancy.
Transport mechanisms:
Gb: FR over ATM
Gn: GTP(UDP)-IP over PPP
Gi: IP over ATM