10.3 Internetworking Solution

For the interworking with other networks, the first thing to consider is the service interworking. Presently, the services provided by the operators include fixed telephone service, IP access service, IP toll service and mobile phone service. In addition, the multimedia video phone service may be a key service of the 3G networks, for which the interworking should be considered.
For the interworking with other networks, the second thing to consider is the allocation of key resources to services, which affects the QoS. For the voice communication, it is required to consider the proper configuration of EC resources while preventing other services from wasting the EC resources.
For the interworking with other networks, it is required to consider the security and reliability of the whole network. It is better for the gateway exchange construction to follow the principles of large capacity and paired construction.
For the interworking with other networks, it is better to access the nearby network to affectively avoid the alternative routes of the mobile subscribers. The remote network may also be accessed for this purpose, which is useful for the charging settlement and the full utilization of EC functions.

10.3.1  Numbering Plan

Now, the numbering plans adopted by the mobile communication include the network numbering plan and the PSTN numbering plan.
The net numbering plan is the numbering plan adopted in current China mobile communication system. Both China Mobile and China Unicom adopt this kind of numbering plan. No further description is given for the network numbering plan in this document.
The PSTN numbering plan is to number the mobile subscribers and equipment with the original PSTN coding scheme. The plan is consistent with the PSTN as to the dialing scheme with a short number and easy availability of the number.
Advantages of the PSTN numbering plan:
l   The types of the numbers are the same, so such original equipment of PSTN as the tandem exchange and the toll exchange may be still adopted by only adding relevant data configurations.
l   Compared with the network numbering plan, its number length is shorter for the convenience of dialing by fixed subscribers.
However, the following problems may occur in this kind of numbering plan:
l   Alternative routes may occur when the called is roaming.
l   After the separation of North Telecom and South Telecom, this plan may obtain different number resources from different regions, or be different from the dialing habits of subscribers. It is not certain whether subscribers can accept such a plan.
The situation of severe competitions among the operators in the China domestic communication market comes into being. The operators build their independent networks, which interconnect via gateway exchanges. The adoption of the PSTN numbering plan will greatly increase the data configuration pressure of some gateway exchanges, such as the international incoming calls or the international roaming.
With MAP routing analysis and call routing analysis, the adoption of the net numbering plan is helpful for the network selection with simple data configurations. It is better to adopt this numbering plan for the call among some international gateway exchanges or some operators, while the PSTN numbering plan is useful to select the geographical position of the call or MAP signaling routes.

10.3.2  Gateway Exchange Solution

1. Requirements for gateway exchange construction

1)      Have such features as large-capacity trunks, powerful signaling processing capacity and abundant voice resources, thus satisfying the development trend and construction requirements of the gateway exchange for “large capacity & few offices”.
2)      Have a variety of interfaces (at least E1 interfaces available) to support diverse services and access modes; and provide ISP with the PRI interface to adapt to the fast development of the Internet.
3)      Have echo cancellation function to reduce the echo resulting from the two-wire/four-wire transformation between mobile calls and fixed calls with the relevant echo canceller configured in the gateway exchange.
4)      Have the built-in SDH to provide the 155M interfaces for implementing the centralized traffic, few office directions and large trunk groups in the gateway exchange, so as to reduce the networking cost, the utilization area of equipment room and the system power consumption.
5)      Have the ability to analyze the calling number of the incoming trunk.
6)      Support flexible networking: For the local network without LSTP or enough LSTP capacity, the gateway exchange may be served also as LSTP; for the mobile network providing abundant services, it is required to support 2M signaling or multi-signaling point technology to satisfy the high capacity requirement of the signaling links; For the exchange with over 4096 circuits in a single office direction, the switching equipment must support multi-signaling point technologies.
7)      Have the SSP/IP function, the CAMEL and upgrading capabilities for the convenience of CAP capability expansion according to the service requirements.

2. Gateway exchange construction solution

The networking of the GMSC is very simple. The GMSC is usually the external interface exchange of an operator. For the outgoing calls, it is to transfer the voice channels out directly; for the incoming calls, it needs to inquire the MSRN of the called, and then connect the call according to the MSRN. At the early stage of construction, generally the GMSC is also served as the VMSC. With the increase of VMSCs, independent GMSC construction gradually begins (if there are more than three VMSCs) to reduce alternative voice channels between these VMSCs.
For the former mobile operators, construction of the mobile gateway exchange is to build new ones and upgrade old gateway exchanges. It is required to consider the networking setting problems of the gateway exchange, that is, the planning of the gateway exchange.
For the operators that own fixed networks, the following solutions are available for the construction of their mobile gateway exchange:
1)      Solution 1: The fixed gateway exchange is integrated with the mobile GMSC by upgrading the fixed network gateway exchange
The following functional upgrading should be carried out for the fixed gateway exchange:
l   Increase MAP functional modules for route enquiry.
l   Increase EC functional modules.
l   Increase the functions of CAMEL3 SSPs.
l   Increase relevant billing functions.
l   Enhance the number analysis function.
If only partial upgrading can be achieved for the current fixed gateway exchange, some local networks should be set separately and some may be integrated. For the office that can be upgraded, it is also necessary to check the current traffic load of the gateway exchange. If the traffic load is heavy, the switch should be checked to find whether capacity expansion can solve the load problem. For this reason, the following requirements should be satisfied before carrying out the upgrading:
l   The current traffic load is light.
l   The current traffic load is heavy but it may be solved through capacity expansion.
l   The equipment vendors ensure the exchange can be upgraded smoothly.
Advantages of the solution:
l   The GW and GMSC integration is in accordance with the network construction trend of large capacity and few offices. Especially at the early stage of networking, the existing resources of the fixed gateway exchange are utilized to carry out efficient networking, thus shortening the construction cycle and reducing the construction cost.
Disadvantages of the solution:
l   It requires to upgrade the fixed gateway exchange, which may affect the existing network.
l   Part of the gateway exchange cannot be upgraded, which makes the networking structure unclear.
2)      Solution 2: Build a new 3G mobile gateway exchange for the fixed network
This solution is to build a new 3G mobile gateway exchange separated from the fixed gateway exchange.

Advantages of the solution:
l   There is no change to the old gateway exchange, no risks from the upgrading, and no impact on the existing fixed network from the traffic load.
l   The networking structure is clear, which is helpful for future network optimization and upgrading.
l   It is helpful for the integrated operators to carry out independent operation management and independent accounting of the mobile and fixed networks.
l   The EC allocation and utilization is clear, which is helpful to reduce the EC costs.
Disadvantages of the solution:
l   At the early stage of construction, it causes the trunk waste; and compared with the integrated gateway exchange solution, it increases the constructions costs and operation costs of the office and its auxiliary equipment.
l   Although the integrated operators have fixed telephone networks, they treat their mobile phone networks the same as those of other mobile operators with two GWs passed, which increases the links of the call and affects the network quality.
Generally, independent accounting is required for each internal network of the integrated operators. At present, the interconnection and accounting of the internal networks are carried out by building their respective gateway exchanges. Thus, each private network needs its own independent gateway exchange, which results in repeated investment and does no good to the unified internal accounting.

10.3.3  Echo Canceller (EC) Configuration Solution

The EC configuration is critical for the interconnection between the 3G mobile network and the PSTN. The EC has the automatic detection function to detect the modern signals and can disable the EC function automatically, so it will not affect the data service functions. However, the equipment cost is fairly high. To configure EC resources where unnecessary will increase the costs of network construction.
EC should be configured as near as possible to the fixed telephone. Generally, the EC can cancel 64 ms echo of the toll calls within about 6500 km. Echo problems occur if the distance is more than this figure.

1. Access mode of the EC

MSC provides two types of echo cancellers: Embedded echo canceller and independent echo canceller.
The embedded echo canceller is based on the concept of exclusive use of resources, that is, one trunk circuit exclusively occupies one echo canceller. When the E1 signals access the MSC, the signal echo cancellation function is implemented.
The independent echo canceller is based on the concept of resources sharing. All the echo cancellers are placed in the ECPOOL (Echo Cancellation Resources Pool). The echo canceller is occupied upon request and then released after use for reoccupation by other connections. The number of independent echo cancellers may be configured flexibly according to the traffic and the number of trunk equipment generating the echo to maximize the resources sharing.
From the perspective of system resources occupation, the ECPOOL should occupy the interface frame slot. For the call connection that requires echo cancellation, if an independent echo canceller is adopted, then four switching timeslots will be occupied in the central switching network: One is connected to the incoming E1 timeslot, one to the outgoing E1 timeslot, and the other two to the echo canceller as shown in the following figure. And if an embedded echo canceller is adopted, then the connection is the same as the common call connection with only two switching timeslots occupied: One is connected to the incoming E1 timeslot, and the other to the outgoing E1 timeslot shown in the following figure. That is, the independent echo canceller realizes the global sharing of echo cancellation resources by occupying more system resources than the embedded echo canceller. In particular, the EC is usually configured on the GMSC equipment with great capacity. The shared EC will greatly reduce the number of E1s that can access the GMSC.

Although the embedded EC can save the switching resources, sometimes the ECs are wasted. For example, the data services (including dial-up access and multimedia calls) are converted with IWF equipment and need no EC to cancel echo any more. In such a case, EC will not affect these services, but it is a waste of the EC processing resources. The embedded EC is suitable for the GSM network, in which there are few such services and little waste of EC resources.
In the other case, PSTN calls the mobile subscriber and then forwards the call to another PSTN subscriber, thus forming an end-to-end PSTN-to-PSTN call and causing a waste of the embedded EC resources.
The adoption of ECPOOL may cause a waste of EC resources, because the TUP protocol and ISUP protocol on the network do not support the EC processing. In this way, the GMSC with ECPOOL resources does not know whether the exchanges in the call route have applied for EC. Incorrect judgment will cause excessive or insufficient EC resources to be requested, thus lowering the communication quality.

2. Configuration of data service channels

The data service, multimedia service and fax service need no EC equipment. PSTN calls the mobile network, and the call is forwarded to the PSTN again without passing the EC equipment. When interconnecting with the PSTN, the data service and the voice service may have their respective trunk groups or office directions, with one for speech call and one for data service or multimedia service. This requires that the GMSC should have the ability to select different routes for the data service and the voice service.

If the data or multimedia traffic is not large, then data services and voice services may not be separated. In the current GSM network, they are usually not separated. However, the 3G multimedia service is considered as a key service with wide application, so it may be separated to some extent.

10.3.4  Routing Mode

The routing mode is considered generally from the perspective of the caller.

1. Access to the nearby network is recommended for internal calls of the operator

Considering the fact of internal calls and all revenues to the same operator, technically the mode of accessing the nearby network is recommended.
The advantage of this mode is that it may effectively reduce the alternative routes of the mobile services. Its disadvantage is that it does no good to the fully play of the EC function when the mobile network calls the toll fixed network.

2. Access to the remote network is recommended for the calls among operators (a subscriber of the local network calls a subscriber of other networks)

In this mode, the toll service adopts its own toll network, thus achieving the maximum benefits during internetwork settlement.

10.3.5  R4 Interworking

The 3G R4 network is characterized by the separation of the MSC server from the MGW and the use of ATM/IP as the transmission bearer. Just like R99, the R4 network needs the interworking of the bearer connections and the conversion of the user plane media stream formats, as well as the echo control and the interworking of control layer protocols. As to the service interworking mode, R99 and R4 are almost the same.
On the whole, the interaction between the 3G R4 core network and the PSTN is completed via the ISUP signaling and trunk signaling. Just like the traditional core network, 3G R4 should be configured with a dedicated gateway exchange to fulfill the interworking requirements as described above, so as to simplify the internetwork interactive connections and reduce the number of interactive points for the convenience of internetwork settlement and interworking resources management. Based on the architecture of separating the 3G R4 core network bearer from control, it is clear that the GMSC server undertakes the signaling interworking in the application layer of the control plane. The signaling bearer interworking is completed by the independent signaling gateway SG equipment or GMSC server, while the interworking function between the bearer connection and the multimedia layer depends on the MGW equipment.

In the traditional mobile network and the PSTN interworking model, the interworking between the signaling control plane and the bearer plane is implemented by the centralized GMSC with large capacity. In the 3G R4 model where bearer is separate from control, the same MSC server with large capacity may control several MGWs, so the interworking may be more flexible according to the MGW capability and the peripheral networking conditions, for instance, the independent gateway MSC server or the integrated MSC server (that combines the local exchange and gateway exchange functions) controls the interworking between the multiple local MGWs that provide TDM interfaces and the local PSTN network.

10.2 Total Network Solution

Lots of factors should be considered in the WCDMA network construction. For example, the operators should carefully consider the following issues: The investment of construction capitals, the utilization of the existing network resources, the planning of the networking form, the smoothness of network upgrading, the implementation difficulty etc.., as described below.

10.2.1  CS Domain Construction Solution

The construction of the CS domain in the core network is always the focus no matter you upgrade the GSM network to the WCDMA network or upgrade the versions of different protocols in the WCDMA network. Presented below is about the solution of establishing a new WCDMA CS domain network.
The network resources of the current operators are abundant: TDM networks, IP networks and ATM networks. The bearer used for networking is decided by the condition of the specific bearer resources.

1. Solution 1: Building the CS network in the R99 protocol mode

When this solution is adopted, the original TDM transmission network can still be used for transmission of the R99 CS domain. As for the gateway and toll tandem equipment, you can upgrade the original GW or establish WCDMA equipment through stacking. The advantages and disadvantages of the two modes are already given in the previous chapter.
With the R99 construction solution, we can make the best of the resources such as transmission resources, gateway office and tandem office, the good compatibility with the original network equipment can be well guaranteed, and voice services can be provided with good QoS. Therefore, this solution is an economical and quick method to introduce the 3G systems.

2. Solution 2: Building the CS network in the R4 protocol mode

In building the CS domain through the R4 network which separates MGW from the MSC Server, you can use ATM/IP in the internal core network for transmission and use the GMSC for the conversion of voice codec as well as the conversion of ATM/IP to TDM.
The advantages of building the network through R4 are given as follows:
l   It makes networking very flexible. You can configure the network capacity flexibly through MGW according to the local traffic and conduct centralized management and configuration through the MSC Server.
l   It helps evolve to the future packet-based network. It is also helpful to save the bandwidth for transmission.
l   With Trfo technology, the voice quality is improved and the voice codec equipment can be saved.
The disadvantages of building the network through R4 are as follows:
l   It is not easy to reuse the original PSTN equipment, because the ATM/IP technology is also used in the transmission of the signaling and it is difficult to use the original signaling network. We need the signaling gateway to interwork with PSTN.
l   With the separate architecture, we need to take the interworking between the MSC Server and MGW into consideration. However, the test of the compatibility may lead to the delay of the network construction.

3. Solution 3: Building the R99 network through R4 (with bearer and control separated)

In building the CS domain through the MGW and MSC Server of R4, the MGW and MSC SERVER are located in the same place when it comes to the construction; and the MSC Server accesses the MGW via the LAN.Other characteristics are the same as that of the R99 network. Therefore, solution 3 has the same advantages and disadvantages as solution 1.
Compared with solution 1, the major advantage of solution 3 is that it facilitates the transition to the R4 architecture as these facilities have been equipped with ATM/IP interfaces and the signaling processing capability for the R4 networking. A gradual transition is available through replacing boards or adding the corresponding MGW to enable the transition from the R99 to the R4.

10.2.2  PS Domain Construction Solution

The construction of the PS domain comprises two aspects: 1) the construction of NEs such as SGSN, GGSN, CG, and DNS; 2) the construction of the GPRS backbone network, namely the construction of the WCDMA PS domain backbone network.

1. Construction of the WCDMA backbone network

There are several construction methods for the WCDMA backbone network:
l   By using the existing IP network.
l   By using the existing ATM network.
l   By using the private line network.
l   By using the above methods for hybrid networking according to the local conditions.
Please note that if we use the existing IP network to build the backbone network, we need VPN and firewall for security, while if we use ATM, the firewall is unnecessary.
Normally in the PS domain of the WCDMA system, we can use the PVC of ATM or IP for bearing on the Gn interface.
l   As for the provincial Gn interface, if we use the ATM network, it is enough to achieve high security without adding a firewall at the Gn interface, while if we use the IP network, the cost is relatively low but extra methods are needed to ensure the security, such as a firewall and IPSec encryption.
l   As for the interconnection with external networks, the IP mode is always needed. However, for the interconnection with the GPRS backbone network, a firewall must be set.

2. Construction of NEs in the PS domain

The NE form of the PS domain and the basic functions are just the same no matter it is the GPRS network, or the R99/R4/R5 network of WCDMA.
As to new operators of WCDMA, they should establish a new WCDMA PS domain network.
As to the operators of the existing GPRS network, there are two solutions for building the PS domain NEs.
Solution 1: Upgrading the existing GPRS network to the WCDMA network:
As the difference between the GPRS network and the WCDMA PS domain network lies in their access networks, the SGSN-related interface modules need to change their Gb interface into the lu-PS interface.
l   SGSN should support the access of the original BSS and the new RAN at the same time.
l   SGSN, GGSN and CG should support the relevant procedures and services of GPRS and WCDMA.
l   It is required to support the integrated billing of GPRS and WCDMA.
l   It is not difficult for upgrading. You can save part of your investment.
l   The original equipment has been tested on the network, so it enjoys higher stability as compared with the new equipment.
l   The existing packet network architecture can be maintained to keep the overall stability of the existing network.
l   It has little impact on the facilities in the original equipment room. We need not reconsider building any new equipment room or placing the facilities.
l   Upgrading the old one may impact the existing network.
l   Service availability after the upgrading varies with the original GPRS platform.
In fact, the difficulty of implementing this solution and the upgrading smoothness depend on the building ability of the GPRS network equipment. If the original GPRS NE is developed on the basis of the GPRS protocol without considering the subsequent transition to WCDMA, it will be very difficult to upgrade the GPRS NEs, especially to upgrade it smoothly. You can do nothing but establish a new suite of WCDMA PS domain equipment. On the contrary, if the original GPRS NEs have powerful functions with good foresight and universal architecture, it can be upgraded smoothly and save a lot of investment.
Solution 2: Establishing a new WCDMA packet network
If the original GPRS network cannot be upgraded smoothly, or it is not worthwhile to upgrade it, the operators can choose to establish a new WCDMA packet network.
The new WCDMA network can coexist with the original GPRS network at the initial stage, but it shall gradually switch the GPRS subscribers to the WCDMA packet network.
The disadvantage of this solution is that investment is needed in building the new equipment. It does not allow us to utilize the original equipment, and we still have to consider building new equipment rooms and placing these equipment.

10.2.3  Signaling Network Solution

1. R99 signaling network solution

1)      Principles for signaling network construction
l   High reliability of the equipment to ensure high security of the signaling network.
l   Supporting dual backup of the network without single point failure to ensure high security of the signaling network.
l   Powerful processing capability to adapt to the expansion of network scale and the growth of services.
l   Low delay to ensure the service connection speed.
l   Even load distribution to evenly plan the flow of load on the signaling network.
The load of the trunk signaling is light, and if each site has a direct signaling link, there will be too many signaling links; therefore, in most regions the signaling link is available only for the interconnection between the TMSC and the STP equipment, while the interaction between the TMSC and other sites is completed via the STP.
Generally, the dual-net dual-plane networking mode is adopted for the signaling network to ensure high security of the signaling network. STP equipment should have powerful processing capability to adapt to the network scale expansion and the service growth. It also should have low delay to ensure the service connection speed. In addition, the signaling links in the 3G mobile networks need to be organized carefully to avoid too many signaling transfer points. At present, the networking modes of the fixed and mobile signaling networks are the same. A direct signaling link should be set between two SPs with a large information volume, especially between the MSC/VLR and the local HLR when the transmission condition permits. There are two SCCP addressing modes: GT and DPC addressing modes. The GT addressing mode is adopted for inter-province networks while the DPC + SSN addressing mode for intra-province networks. In this way, the work amount of GT translation of the STP equipment can be greatly reduced.
2)      Difference between the mobile signaling network and the fixed signaling network
The hierarchical signaling network is suitable for the fixed signaling network. At present, for the fixed signaling network a pairs of independent HSTPs are generally set in the provincial capital to form a dual-plane mesh network. For each of the local networks, a pair of LSTPs are available to transfer the PSTN signaling and the intelligent network information. An important application of the LSTP is to complete the inter-office conversion from mesh signaling network to hierarchical signaling network, which greatly reduces the amount of direct signaling links and improves the reliability. However, the mobile network features high capacity of local SPs, few SP sites and large signaling flow between local SPs, so it is better to adopt direct links between SPs, and in that case the signaling link convergence functionality of STP is not applicable to the mobile network, rather, the signaling networking of mesh topology is recommended for the local mobile network.
If the STP network of PSTN directly serves as the mobile signaling network, it functionally makes no difference except the following issues:
l   The mobile communication system use the GT addressing mode largely and the processing of GT code is in the SCCP layer, so the STP network of PSTN need to be upgraded if not support SCCP function.
l   The signaling of the local mobile network usually adopts direct links and the signaling to other local networks should pass the STP. However, the LSTP of the current PSTN is only used to transfer the signaling of the local network, thus all the signaling information to the LSTP should be transferred from the HSTP, which increases the load of the HSTP (that is, it has to transfer both the domestic roaming signaling and the intra-area signaling).
l   There exist multiple intra-area signaling transfers, which causes time delay of the connection and affects the QoS to some extent.
3)      Signaling network construction solution
Two networking solutions are available for the mobile signaling network:
Solution 1: The mobile equipment of each local network is only connected with the LSTP, which is used to transfer the intra-province signaling and hand over the inter-province signaling to the HSTP.
Advantages of solution 1 are simple structure and easy capacity expansion. Its disadvantage is also obvious: The inter-province signaling passes LSTP and this increases the load of the LSTP and the time-delay of inter-province roaming and call signaling processing.
Solution 2: The mobile equipment of each local network is connected with both the LSTP and the HSTP. The mobile equipment of the local networks recognize the intra-province signaling and the inter-province signaling to forward the signaling to different STPs for processing.
The advantage of solution 2 is that it reduces the load of LSTP and the signaling processing delay. Its disadvantage is that the capacity expansion is not convenient. With the increase of the network nodes, the utilization of the signaling links to the HSTP will be inefficient to some extent.
To fully capitalize on the existing HSTP resources and create new LSTPs to transfer the inter-province signaling, solution 1 with minimum changes may be adopted for networking.
4)      Introducing high speed signaling links to the mobile network
At present, the SS7 network of the TDM-based mobile network aims to providing transmission bandwidth and reliability for the signaling network and further improving the signaling network management functions. From the perspective of network development, the capability of 3G signaling networks is much greater than that of GSM. That is, with the expansion of the signaling network capacity and scale, and the improvement of the signaling network reliability, the signaling protocol needs to be upgraded further, so as to enrich the services provided to the subscribers.
At present, the SS7 mobile signaling network generally adopts 64Kbit/s signaling links. Restricted by the SS7 protocol, the maximum signaling bandwidth between the nodes of the mobile signaling network is only 1024K (64K´16), which cannot satisfy the requirements for the signaling bandwidth between the MSC and the HLR. The mobile signaling network needs to adopt 2Mbit/s high speed signaling links as soon as possible to increase the bandwidth between the nodes.
To further improve the network reliability, the transmission paths of the signaling network should be dispersed as much as possible and the signaling network management functions should be enhanced.

2. R4 signaling network construction

In the 3G R4 stage, the bearer of call-independent signaling can only be the TDM-based SS7. It may be upgraded to the optional IP-based SIGTRAN or still adopt the TDM-based SS7. According to the 3GPP specifications, the SCCP/TCAP-based MAP/CAP signaling must adopt the M3UA/SCTP SIGTRAN bearer mode, and M3UA provides the upper SCCP with the primitive interface completely equal to the MTP3 protocol, so no matter the 3G R4 signaling bearer is TDM/SS7 or IP/SIGTAN, it is required to construct a hierarchical private roaming signaling network with network layer signaling transfer capability, that is, the STP network.

If the hierarchical structure matches and the TMSC server processing delay is enough, the IP STP equipment may be integrated with the TMSC server equipment physically, so as to effectively reduce the networking investment of the operators.

10. WCDMA Network Solution 10.1 Overview of WCDMA Evolution

10.1.1  Overview of Standard Evolution

The WCDMA technology has gone through R99/R4/R5/R6 stages since it comes into being. The R99 protocol was functionally frozen in March 2000 (December 1999 in 3GPP official document) and almost became mature after two years of improvement. The R4 protocol was functionally frozen in March 2001. The R5 protocol was functionally frozen in March 2002 (with some functions frozen in June), and the R6 protocol is estimated to finalize its functions in December 2004.

Compared with GSM and GPRS networks, the most significant change of the WCDMA system is the change of the radio network. In the WCDMA network, the Radio Access Network (RAN) is used to replace the Base Station Subsystem (BSS).
The WCDMA core network in the R99 version can be regarded as a combination of GSM and GPRS core networks in terms of networking, namely, the R99 core network is classified into the Circuit Switched (CS) domain and the Packet Switched (PS) domain. The architectures of the CS domain and the GSM core network are basically the same, so are the architectures of the PS domain and the GPRS core network.
In contrast to the R99 version, the biggest change of the R4 core network is that the network element function of MSC in the CS domain of the R99 core network is fulfilled by the MSC Server and MGW in the R4 version, where the MSC Server processes the signaling while the MGW processes the voice. There is no change in PS domain. For details, refer to the architecture description in Chapter 3.
The core network using the R4 protocol has two networking modes: TDM and IP. When the TDM mode is adopted, the R4 network planning and construction are identical with that of R99 to a large degree. For example, in constructing tandem and signaling networks, many considerations are just the same. When the IP mode is adopted, the R4 network planning and construction are largely different from that of R99.
In contrast to the R4 version, an IMS (IP Multimedia Subsystem) domain is added in the R5 core network, together with the corresponding equipment and interfaces, but there is nearly no change in the network structure of CS and PS domains. Meanwhile, some equipment functions are upgraded due to the enhancement of the network functions.

The IMS domain is overlaid with the original PS/CS domain. It is used to control the subscribers’ services. Subscribers can use all kinds of access techniques of the PS/CS domain to access the IMS domain. In the future, new services based on the 3GPP R5 IMS domain will have nothing to do with the access techniques that subscribers adopt. At that time, no matter what technique is adopted, the service needs to be developed only once by the service developers.
The IMS domain adopts SIP as the basic session protocol. As the SIP is used to unify the session model of voice/data services, the IMS domain provides more flexible and simple support to multimedia services.
It also provides abundant service development interfaces. Operators can even provide open session messages to the trusted service providers. Therefore, the services will become even more open and flexible.

10.1.2  Network Construction Solutions

It becomes a persisting issue for operators and equipment providers to construct a WCDMA network with powerful functions and stable performance and also to consider the factors ranging from engineering technical difficulties, capital investment and the network compatibility to evolution.

1. WCDMA network construction

For operators, there are two solutions to building a WCSDMA network: Upgrading the old one and establishing a new one.
As to new mobile operators, they will usually choose to establish a new WCDMA network. For GSM and GPRS operators, they need to consider issues such as the network compatibility and roaming. However, gradually they will also choose to establish a new one. The following is a comparison between these two solutions.
Solution 1: Upgrading from GSM to the WCDMA R99 network:
This solution will turn the 2G GSM network into a network with 2G/3G coexistence by upgrading MSC, HLR and GMSC of the GSM network into 3G MSC, 3G HLR, 3G GMSC.
l   The upgraded MSC should support access of WCDMA and GSM radio network equipment at the same time.
l   The MSC should support the lu interface based on ATM and the A interface based on TDM.
l   Support AMR voice, H.324M multimedia calls and TC equipment upgrading.
l   Support double authentication via GSM/WCDMA, 3G MAP, internal handover and compatibility.
l   Support CAMEL3 VAS and the open capability of MSC OSA.
l   Support the integration of 2G/3G charging systems.
Advantages of the upgrade solution:
l   Minimize inter-MSC handover at the initial stage.
Disadvantages of the upgrade solution:
l   Have a big impact on the stability of the existing GSM network.
l   Require a heavy workload and increase difficulties in the compatibility test.
l   Cause the decrease of the existing network capacity, since 3G services are even more complex and demand better processing capability.
l   Affect the service provisioning and cause difficulties in smooth evolution, especially the provisioning of NGN technology, IN service and the third-party applications.
l   Require high costs of upgrading and large investment. The 3G MSC adopts the highly-advanced technical platform with large volume and high integration. Its construction cost is lower than that of the 2G MSC; therefore, it would be not worthwhile to upgrade the small-capacity 2G MSC.
l   The upgrading of the original charging system will have a big impact on the network operation. The interworking and compatibility of multiple manufacturers and nationwide roaming need to be verified.
This solution will give a full-scale evaluation to the problems such as the impact on the stability of the existing network, the continuity of service capabilities, service processing capacity, capacity, level of integration and whether the seamless transition of GSM-R99-R4-R5-R6 is available after the upgrading. It is hard to know whether the traditional architecture of 2G switches meets the above requirements, therefore, it is also impossible to know the investment utilization rate of the original GSM equipment.
Solution 2: Establishing a new WCDMA network
This solution is to establish a new 3G MSC and a new 3G RAN based on the original GSM network. The GSM and WCDMA networks will coexist in a certain period of time with service interoperability and the WCDMA network will gradually replace the GSM network.
Requirements for the core network equipment:
l   Establish a new WCDMA RAN and a new MSC to support CAMEL2, 3 IN service and the access of 2G/3G networks simultaneously.
l   The existing SGSN and GGSN have relatively small traffic and can be upgrade into 3G equipment. They should be compatible with GPRS access and applications and support CAMEL3 IN services.
l   For the WCDMA HLR, both solutions (upgrading the old one and building a new one) are available. As there are few HLRs, the upgrading is quite easy.
l   The backbone and core networks of TMSC and GMSC should be shared. GMSC upgrading can be disregarded at the initial stage but later they may support 3G MAP and CAP protocols and the trigger of 3G IN services.
l   The signaling network equipment LSTP/HSTP of 2G SS7 should be shared.
l   The 2G MSC should stop capacity expansion and BSS should be connected to the new 3G MSC.
Advantages of the solution for building a new WCDMA network:
l   As a new network, it is easier for the entire network to plan the resources and configurations uniformly in order to have a clear network architecture.
l   With large capacity and less offices, it can simplify the network architecture to facilitate centralized maintenance and management.
l   The test becomes more convenient and comprehensive.
l   Have a small impact on the existing network during the implementation.
Disadvantages of the solution for building a new WCDMA network:
l   There may be coordination problems with the handover between 2G and 3G systems. If the manufacturers of the existing network do not set obstacles deliberately, it is easy to solve this problem.

2. Evolution from R99 to R4

Evolving from the R99 network to the R4 network, you need replace the MSC NE of R99 with the MSC Server and MGW of R4 from the perspective of the protocol layer. However, in terms of the specific implementation, the construction solution of the R99 network at the initial stage is closely related with the amount of investment needed for evolution and the difficulty of upgrading.
1)      Solution 1: Upgrading the R99 network to the R4 network
If the MSC NE of the R99 network has already taken into account the idea of processing the signaling and voice separately (that is, the MSC based on the design philosophy of separating bearer from control is adopted to build the R99 network), the evolution will be very easy no matter upgrading from R99 to the R4 of TDM networking, or even to the R4 of IP networking and only the equipment needs to be upgraded.
l   The original R99 should be capable of separating bearer from control to facilitate the transition to the R4 architecture.
l   The upgraded equipment should support all the signaling interfaces of the R4 network.
Following is a comparison between the advantages and disadvantages of this solution:
l   Advantages: It is easy to upgrade the network. You need only upgrade the original equipment instead of purchasing any MSC Server or MGW. It is also easy to upgrade the ATM/IP-based network. The most important thing is that it brings very big flexibility to the construction and planning of the network.
l   Disadvantages: The R99 network at the initial stage should meet higher requirements. The R99 network upgraded directly from the GSM network probably will not meet these requirements. In this regard, you have to establish a new MSC Server and MGW.
2)      Solution 2: Establishing a new R4 network
If the MSC of the R99 network fails to take into account the idea of processing the signaling and voice separately at the later stage (or if this R99 network is upgraded directly from the GSM network), probably the bearer and control modules cannot be well separated. In that case, the evolution from R99 to R4 cannot be completed through upgrading the MSC, or we can say it is not worthwhile. In this regard, we have to purchase additional MSC Server and MGW.
Following is a comparison between the advantages and disadvantages of this solution:
l   Advantages: The new network has powerful functions and is capable of smooth transition to the later all-IP networking as well as R5.
l   Disadvantages: It is required to establish a new network, which makes the cost higher than the upgrade solution.
To summarize, it is a continuous evolution process from GSM to R99 and then from R99 to R4 in terms of the whole version evolution. During the WCDMA network construction at the earlier stage, if the solution to upgrading the GSM network to the R99 network is adopted, you need rebuild the MSC Server and MGW NE in upgrading from R99 to R4. However, if you establish new R99 network with the MSC that is able to separate bearer from control, the cost at the initial stage may be higher than that of the upgrade solution, but it is conducive to the later smooth transition to R4 and R5. Therefore, the comprehensive investment efficiency of the solution for establishing a new R4 network will be much higher.

3. Evolution from R4 to R5

In R5, an IMS domain is added. The logic block diagram of the IMS domain is, where the Go interface is interconnected with the GGSN while the Mb interface is used for the external network to access the IMS domain.

The functional and logic entities of the IMS include BGCF, CSCF, MGCF, IM-MGW, MRFC, MRFP, HSS and SLF. Some logic entities can be evolved from the R4 function entities while others must be added.

Therefore, it is better to establish new equipment of the IMS domain for evolution from R4 to R5 and upgrade the functions of the equipment in the original CS and PS domains. 

9.2 Billing Principles of WCDMA PS Domain

9.2.1  WCDMA PS Domain Billing System Architecture

Billing system of PS domain consists of three parts: SGSN/GGSN billing module, CG and billing center. There is a standard Ga interface with GTP’ protocol between SGSN/GGSN and CG. Between CG and billing center, bill data are transported with FTP/FTAM file transporting protocol.
The functions of each part are described as follows:
SGSN and GGSN: Generating billing data.
l   Collecting the billing data of SGSN and GGSN.
l   Saving for a long time and executing some pre-processing work, such as integrating and sorting.
l   Transporting the collected billing data to the billing center.
Billing System (billing center) : Processing billing data and generating final bill.
Note: In Ga interface, the billing data is called CDR. CDR that is Call Detail Record.

9.2.2  Billing Data Generated by GSN

1. Billing data

There are 5 types of CDRs generated in 3G billing:
l   M-CDR: Generated by SGSN, used for recording billing information of mobility management for mobile phone. It can be configured whether to generate this bill
or not, not required normally.
l   S-SMO-CDR: Generated by SGSN, used for recording billing information of SM origination.
l   S-SMT-CDR: Generated by SGSN, used for recording billing information of SM termination.
l   S-CDR: Generated by SGSN.
l   G-CDR: Generated by GGSN.
As for the same PDP course, there are 2 types of CDRs generated, S-CDR and G-CDR. Billing center usually calculates the final charge according to G-CDR, while S-CDR is mainly used for statistics.
CDR generated in SGSN and GGSN mainly records the following information:
l   Radio resource utilization
l   Duration
l   GPRS resource utilization
l   Originator and terminator
l   External data network utilization
l   Mobile terminal location.

2. Integration of partial CDRs

One PDP context may be corresponding to multiple partial CDRs, which are generated because of:
l   Data amount limitation
l   Time limitation
l   Changed billing condition (such as tariff change).
Because in the bill finally sent to subscriber, there is only one bill in one PDP context, all partial CDRs in one PDP context must be integrated. Integration of partial CDRs is carried out in 2 steps: The integration in the first step is carried out by CGF. This can reduce the requirement of bandwidth between CGF and billing center and alleviate the processing operation of billing center. For this reason or that, the integration in this step may be incomplete. The second step is carried out by billing center, where those CDRs that are not integrated completely in CGF will be integrated, resulting in final CDRs.
As for each PDP context, GGSN generates one exclusive C-ID. According to C-ID+GGSN address, it is possible to know whether two partial CDRs belong to the same PDP context.
As for G-CDR, all partial CDRs with the same C-ID+GGSN address must be integrated.
As for S-CDR, all partial CDRs with the same C-ID+GGSN address and +SGSN address must be integrated.

9.2.3  CGF

CGF offers a mechanism to send the billing information generated by GGSN and SGSN to designated billing center. CGF is an abstract functional concept, and CG is a specific implementation of CGF. The specific implementation of CG differs with manufacturers.
CGF must offer the following functions:
l   Collecting CDR from SGSN and GGSN
l   Providing long-time reliable storage of CDR
l   Transporting CDR to billing center.
In addition, to reduce the transmission amount between CGF and billing center, CGF should provide some integration function of partial CDRs to minimize the amount of CDRs transmitted to billing center, thus lightening the bandwidth requirement for the billing center.
CGF must offer high degree of reliability and redundancy. One GSN should be corresponding to multiple CGFs in different levels of priority. When CGF in higher priority can not process the communication with GSN, GSN will redirect it to CGF in lower priority.

9.2.4  Billing Center

Billing center has the following main functions:
l   Collecting the CDRs on CGF
l   Figuring out charge according to bill
l   Fully Integrating CDRs
l   Proofreading bills.

9.2.5  GTP' Protocol

GTP' protocol is a communication protocol between GSN and CGF and between different CGFs. It is an application layer protocol. The bottom protocol stack adopts UDP/TCP and IP.

The major functions of GTP' protocol are as follows:
l   Transporting CDRs from GSN to CGF
l   Redirecting CDRs to another CGF
l   Detecting the communication failure between CGF and GSN
l   Recommending another available CGF to GSN in case of failure in one CGF
l   Preventing duplicate CDRs.