1 Introduction

Macro coverage and indoor coverage could not be optimized during the early phase of TD core network construction. And there are many 2G/3G borders in the same area when 2G/3G interoperability is enabled on the network, especially in the same location area (LA). As a result, a great number of location updates and paging signals occur between 2G and 3G networks [1–6].

Because the users are unreachable during location update, the voice call completion rate of the TD network decreases badly. It is possible to set location area codes (LACs) of 2G and 3G networks uniformly in areas in which this problem is rather severe to enable co-routing. In this manner, terminals in idle state do not perform immediate routing area update after cell reselection within the same routing area. Terminals initiate routing area update only when the network side initiates paging or terminals initiate services. This can reduce routing area update signaling caused by reselection. In the actual networking [7–9], however, when the core network performs paging based on LACs, paging messages are delivered over the 2G and 3G networks concurrently. Extra invalid paging messages bring paging congestion on the 2G network and radio side of the TD network, therefore affecting the paging success ratio [10, 11].

As a result, in this paper we determine to adopt precise paging mode based on co-LAC for 2G/3G to meet network requirements and improve user experience. We make an analysis of the reduction of paging signals and update messages by path diversity. And the implementation is also done, where the results prove the efficiency of our proposal.

2 Concept and principle

2.1 Concept of LAC

On a cell-based mobile communications network, a great number of base transceiver stations (BTSs) are deployed, and mobile stations are not fixed. When mobile users are in the service area regardless of where they move, the mobile communications network must perform the switching control function to implement location update (registration), cross-cell handover, and automatic roaming. The following figure shows the definition of areas in a GSM network.

An LA refers to an area in which mobile stations can move freely without location updates. An LA can consist of one or several cells (or BTS areas). To call a mobile station, all the BTSs in the same LA can send paging signals concurrently.

The identity of an LA is referred to as location area identity (LAI) in Fig. 1. An LAI consists of three parts: mobile country code (MCC), which is of three numbers and identifies the country to which a mobile user belongs; mobile network code (MNC), which identifies the mobile network to which a mobile user belongs; LAC, which identifies an LA on a GSM Public Land Mobile Network (PLMN). The LAC is the most important portion of an LAI. Therefore, an LA is often referred to as an LAC, and co-LAC means co-LA.

Fig. 1
figure 1

Areas in a GSM network

Full size image

2.2 Concept of location update

Due to MS mobility, it is important to locate an MS. Only after the current location of an MS is known, connection to the called MS can be set up.

Roaming refers to the feature that a mobile user requests changing of the connection to the cell and network during movement. Location update refers to changing of an LA during roaming and LA confirmation. A mobile switching center (MSC) does not need to be informed of movement within the same LA but movement between cells of different LAs. Location update is primarily composed of the following:

When an MS detects that its LAI is to be updated, the MS sets up a connection with the MSC/visitor location register (VLR) proactively. Then, the MS sends a request, and data on the VLR is updated. If the LAI belongs to different MSCs/VLRs, data on the home location register (HLR) also needs to be updated. After update, the connection between the MS and the BTS is released (Fig. 2).

Fig. 2
figure 2

Location Update Procedures

Full size image

Users are unreachable during location update (a mobile phone cannot listen on paging channels during location update. Therefore, the mobile phone cannot respond to paging requests), that is, users cannot be called during location update. Therefore, the call completion rate of them is greatly affected when their locations are frequently updated. In addition, communications are badly affected.

3 Problem description and theoretical analysis

3.1 Problems analysis of update message

  1. 1)

    Different LACs for 2G/3G

As shown in the preceding figure, in the scenario of different LACs for 2G/3G, the chance of TD-to-GSM cell reselection is limited in the area covered by the TD network continuously. Therefore, location update messages for the GSM network do not increase significantly. TD cells are preferred. Therefore, the chance of GSM-to-TD cell reselection is also limited. Consequently, location update messages for the TD network do not increase significantly, either (Fig. 3).

Fig. 3
figure 3

Different LACs for 2G/3G

Full size image

At the coverage edge of the TD network, dual-mode terminal users who enter TD network coverage all initiate location update and register with the TD network because TD cells are preferred. Users who leave the TD network need to initiate location update on the GSM network and register with the GSM network. Compared with GSM and TD networks, location updates increase greatly at the TD network border.

The GSM network provides an extensive coverage, and GSM users seldom need to switch to the TD network. During the early phase of the TD network, there is a big gap between the coverage of the TD network and that of the GSM network, and TD users often need to reselect cells or switch to the GSM network. In this case, location updates occur. Especially in the same geographical area where different LAs are used for GSM and TD, TD users are more likely to reselect cells or switch to the GSM network. Therefore, a great number of location updates occur.

Here, TD cells mean the mobile communication cells covered by TD-SCDMA (Time Division Synchronous Code Division Multiple Access). TD-SCDMA is a 3Generation mobile communication specification which is approved by ITU (The International Telecommunication Union). TD users mean the mobile phone users who are served by TD-SCDMA.

Paging to users is performed on the basis of LAs on the GSM or TD network. LAs are set separately. Therefore, the TD network is responsible for only paging users who really register with the TD network. Users are unreachable during location update. A mobile phone cannot respond to paging messages during location update. Therefore, in the preceding scenario in which location updates increase, the voice call completion rate of the TD network decreases badly, and user experience is also affected. Therefore, the solution of co-LAC for 2G/3G needs to be introduced to solve this problem.

  1. 2)

    Co-LAC for 2G/3G

See the preceding figure. At the coverage edge and areas without coverage or weak coverage of the TD network, dual-mode terminal users who enter the TD network coverage initiate inter-RAT (Radio Access Technology) cell reselection and select the TD network. This is because the TD network is preferred. The core network, however, is not informed of the reselection. Users who leave the TD network also initiate inter-RAT cell reselection and reselect the GSM network. The core network is not informed of the reselection either.

Advantages of co-LAC: When a user performs cell reselection between 2G and 3G wireless systems, the user does not need to initiate location update to the core network. This is because the LAC does not change and the core network does not need to be informed of LAI change. In addition, the delay of system reselection is very short. Therefore, the possibility that paging and system reselection (including location update) occur concurrently can be minimized. In this manner, the paging success ratio increases and the completion rate of calls can be improved (Fig. 4).

Fig. 4
figure 4

Co-LAC for 2G/3G

Full size image

There are three types of location updates: normal location updates (cross-LA), periodical location updates, and International Mobile Subscriber Identity (IMSI) attachment (corresponding to the process of terminal power-on). In the case of co-LAC for GSM and TD networks, the number of periodical location updates and IMSI attachments do not change greatly. This is because mobile phones complete these two types of location updates within their own networks. The major change is that normal location updates decrease. This is because in the same LAC, location updates do not occur just because a TD user reselects cells or switches to the GSM network due to weak signals or insufficient coverage. This also excludes the possibility of call drop due to location update. In the early phase of TD network construction, this can greatly improve the service experience of TD users.

Because the LAC is shared, when the core network pages a piece of user equipment (UE), the core network needs to perform paging over the entire LA, including the coverage areas of both TD and GSM networks.

The GSM network needs to serve users who register with the TD network. Although these paging messages are not responded to, they occupy the paging channels of the GSM network. Considering the small number of users in the early phase of the TD network, the impact on the GSM network is limited after co-LAC deployment. With the increase in the number of TD users, the paging capability of the GSM network may decrease inevitably.

The TD network needs to serve users who register with the GSM network. Although these paging messages are not responded to, they occupy the paging channels of the TD network. The impact to the TD network is great because the number of GSM users is multiple times that of the TD network. To prevent invalid paging in the scenario of co-LAC, the solution of precise paging for co-LAC needs to be introduced.

  1. 3)

    Position Update Message Reduction

When the 2G and TD are located in different areas, the overhead O 1 will be caused.

$$ {O_1} = \sum\limits_{i = 1}^3 {{f_i} \cdot {t_i}} $$
(1)

Here, t 1 means the data size for normal location update, t 2 denotes the data size for periodical update and t 3 shows the data size of IMSI message. And f 1, f 2 and f 3 show the corresponding frequencies.

If the 2G and TD are located in the same area, the overhead O 2 will be caused. Here, \( f_{i}^{\prime } \) denotes the corresponding frequencies in this situation.

$$ {{O}_{2}} = \sum\limits_{{i = 1}}^{3} {f_{i}^{\prime }\cdot {{t}_{i}}} $$
(2)
$$ \left\{ {\begin{array}{*{20}{c}} {f_{1}^{\prime } < < {{f}_{1}}} \\ {f_{2}^{\prime } = {{f}_{2}}} \\ {f_{3}^{\prime } = {{f}_{3}}} \\ \end{array} } \right. $$
(3)

Then we can get,

$$ {O_2} << {O_1} $$
(4)

So the traffic can be reduced more.

3.2 Problems analysis of paging signal

  1. 1)

    Paging to both BSC and RNC

When a mobile phone performs location update in a co-LAC area, the core network obtains parameters of the mobile phone, including classmark1 and classmark3, and saves the parameters to the data area of the VLR.

During paging delivery in the LA for the mobile phone, the MSC determines whether the mobile phone is a GSM-only one. If yes, the paging is sent to only the BSC in Figs. 5 and 6.

Fig. 5
figure 5

Conventional method for paging signal

Full size image
Fig. 6
figure 6

Proposal to determine the type of a mobile phone

Full size image

If the mobile phone is not a GSM-only one or the terminal type is not reported, the MSC sends paging to both the BSC and the RNC (Radio Network Controller) concurrently in Fig. 6.

During location update, the MSC obtains the revision level from classmark1 in the location update message sent from the mobile phone. revision level is a two-bit parameter. When its value is 01 (GSM PHASE 1) or 10 (GSM PHASE 2), the mobile phone is a GSM-only one.

When its value is not 01 (GSM PHASE 1) or 10 (GSM PHASE 2), you can determine the mobile phone type based on the following parameters in classmark3 reported by the mobile phone. If the mobile phone does not report classmark 3, the MSC can send the classmark update message to obtain this parameter proactively.

  • UMTS FDD (Universal Mobile Telecommunications System Frequency Division Duplexing) Radio Access Technology Capability

  • CDMA (Code Division Multiple Access) 2000 Radio Access Technology Capability

  • UMTS (Universal Mobile Telecommunications System) 3.84 Mcps (M codes per second) TDD (Time Division Duplexing) Radio Access Technology Capability

  • UMTS (Universal Mobile Telecommunications System) 1.28 Mcps (M codes per second) TDD (Time Division Duplexing) Radio Access Technology Capability

The preceding parameters represent a special 3G radio access technology. If the preceding bits are 1, the mobile phone supports this radio access technology. If the preceding bits are all 0, the mobile phone is a GSM-only one. Otherwise, the mobile phone is not a GSM-only one.

  1. 2)

    Analysis of Paging Signal Reduction

For the conventional method, if a call (ith, i∊[1,…I]) arrives in the MSC, the system will select two paths to send the paging signal to both BSC and RNC. Assume that data size of the ith paging signal is d i , then the total data size of paging signal becomes,

$$ \sum\limits_i {{d_i} \cdot [f(PS,2G) + f(PS,3G)]} $$
(5)

Where, f(PS,2G) and f(PS,3G) are defined as the path functions of BSC and RNC, respectively. Since both the paths of BSC and RNC are selected, we can obtain.

$$ \left\{ {\matrix{ {f(PS,2G) = 1} \\ {f(PS,3G) = 1} \\ }<!end array> } \right. $$
(6)

The total data size becomes,

$$ 2\sum\limits_i {{d_i}} $$
(7)

In our proposal, if the MSC realizes the mobile phone is a GSM mode, only the path of BSC will be selected as the path to send the paging signal, which means,

$$ \left\{ {\matrix{ {{f^\prime }\left( {PS,2G} \right) = 1} \\ {{f^\prime }\left( {PS,3G} \right) = 0} \\ }<!end array> } \right. $$
(8)

Otherwise, if the mobile phone is a both GSM and UMTS mode, both the paths of BSC and RNC will be selected to send the paging signal. That is to say:

$$ \left\{ {\matrix{ {f\prime \prime \left( {PS,2G} \right) = 1} \\ {f\prime \prime \left( {PS,3G} \right) = 1} \\ }<!end array> } \right. $$
(9)

Assume that there are i’ mobile phone are GSM mode within the total N calls, then we can get the total data size that the paths need to process as follows,

The above analysis proves that our proposal can reduce the data amount (traffic) by selective path.

$$ \begin{array}{*{20}{c}} {\sum\limits_{i}^{{{{i}^{\prime }}}} {{{d}_{i}}\cdot \left[ {{{f}^{\prime }}\left( {PS,2G} \right) + {{f}^{\prime }}(PS,3G)} \right]} } \\ { + \sum\limits_{{{{i}^{\prime }} + 1}}^{I} {{{d}_{i}}} \cdot \left[ {{{f}^{{\prime \prime }}}\left( {PS,2G} \right) + {{f}^{{\prime \prime }}}\left( {PS,3G} \right)} \right]} \\ { = \sum\limits_{1}^{{{{i}^{\prime }}}} {{{d}_{i}}} + 2\sum\limits_{{{{i}^{\prime }} + 1}}^{I} {{{d}_{i}} = 2} \sum\limits_{1}^{I} {{{d}_{i}}} - \sum\limits_{1}^{{{{i}^{\prime }}}} {{{d}_{i}}} < 2\sum\limits_{1}^{I} {{{d}_{i}}} } \\ \end{array} $$
(10)

4 Implementation results

4.1 Tests and verification for various scenarios

To verify whether the quality of network can be improved, we carry out the following tests to evaluate the effects and changes, when the proposal is used. For analyzing the detailed performance of our proposal, we use some metrics including handover success rate and network completion rates.

  • Regression test for typical services after co-LAC deployment to test the validity and consistency of each typical service

  • Test of the precise paging function for the core network

  • Change in delay of 2G/3G interoperability before and after co-LAC deployment

  • Compatibility of PBP parameters for terminals, and test of TD paging capacity

  • Effect of co-LAC configuration on the ARD function of the core network

  • Effect of co-LAC configuration on co-cell ID for 2G/3G

Note that the LA of a TD network is greater than that of a GSM network. LA division refers to dividing a large LA into several smaller LAs so that the LA of the TD network is of the similar size with that of the GSM network.

4.2 Detailed results

The success rates of entire-network handover of MSC pool in four scenarios within 6 days are all about 94 %. Co-LAC does not affect the success rate of location handover.

There are four scenarios. Scenario 1: Different LAC with TD LAC not being divided; Scenario 2:Co-LAC with TD LAC not being divided; Scenario 3: Different LAC with TD LAC being divided; Scenario 4: Co-LAC with TD LAC being divided. Both RNC and BSC share in the same MSC in the real network, each BSC or RNC has been respectively controlled by MSC in MSC pool when MSCs operate in the MSC pool. So the handover success rate of MSC pool needs to be estimated in MSC pool. Here, we consider the different cases of handover including handover occurred inter-BSC, handover occurred inter-RNC, and handover occurred inter-MSC (Figs. 7 and 8).

Fig. 7
figure 7

Handover success rate of MSC pool within 6 days

Full size image
Fig. 8
figure 8

Network completion rates of the MSC pool within 6 days

Full size image

When a 2G/3G customer attempts to call another 2G/3G customer, the process for speaking set up successfully by two of them at last is named network completion. As a result, the network completion rates of the MSC pool is collected and analyzed in order to assess whether Co-LAC can affect the network successful completion. The network completion rates of the MSC pool within 6 days are all about 99 %. Co-LAC does not affect the network completion rate.

The reason for analyzing handover times is that too many handovers can cause network successful completion rate to be decreased. Therefore, it is very important for us to focus on handover times in the target region. Figure 9 shows Co-LAC does not impact negatively handover times by analyzing data.

Fig. 9
figure 9

Handover times in the target region in four scenarios within 6 days

Full size image

Handover times in the target region in four scenarios within 6 days are collected. The average value is 595699 times in scenario 1, 595880 times in scenario 2, 600011 times in scenario 3, and 593429 times in scenario 4. All the values are around 590000. There is no significant variance in the values of different scenarios. Co-LAC does not affect handover times.

5 Conclusions

The proposed precise paging technology can eliminate the bottleneck of co-LAC. The precise paging technology has been implemented. According to actual test, when precise paging is started in co-LAC, the paging success rate of the live network is not affected. The study of situations before and after precise paging is started shows that precise paging can eliminate invalid paging messages of the 2G network upon the 3G network and realize the expected target. Invalid paging messages decrease from 78432 to 705, and the ratio of valid paging increases to 48.23 % from 0.47 %.