Analysis on PHS Network Optimization of DSP Dual-mode Mobile Phone

In this article, we mainly discuss the optimization of PHS network (PHS, P network) and the impact of dual-mode mobile phones on PHS network. Wireless network optimization includes the optimization of terminals, base stations and core networks. GSM (G network) and 3G (C network, including 2G IS-95) have relatively complete network optimization solutions.

At present, there are multiple networks coexisting in China, such as GSM, PHS, and CDMA. In order to make full use of these network resources, multi-mode mobile phones have come into being. Multi-mode mobile phones based on zero intermediate frequency (Direct Conversion) radio frequency (RF) technology and DSP digital baseband processors provide powerful capabilities to achieve high cost performance and reduce the probability of wireless communication failure caused by overload of a single network. In addition, the Xiaolingtong dual-mode mobile phone uses DSP technology to complete the coordinated optimization of the Xiaolingtong network at a relatively low cost, improving the instantaneous load capacity and service quality of Xiaolingtong. The network optimization of Xiaolingtong has not been proposed for a long time, but it is under tremendous pressure.

Problems in Xiaolingtong network optimization
Xiaolingtong has the advantages of low power consumption, low radiation, and low tariffs, and has won considerable favor in the low-end market. However, Xiaolingtong also has some defects, such as low transmission power, severe scattering, and fast signal fading. In response to these shortcomings, the PHS network adopts micro-cellular technology. One of the additional benefits of micro-cellular technology is that it can theoretically carry more users. However, micro-cellular technology requires the establishment of more base stations, especially in urban areas with complex environments and large traffic. There are three factors that affect the quality of Xiaolingtong service, namely terminals (mobile phones), base stations, and core networks. We can regard them together as components of network optimization. Terminals and base stations are the optimization of the wireless part, and the core network is the wired part. Optimizing the wireless part can achieve twice the result with half the effort in improving the quality of system service. Unlike GSM network optimization (in the GSM network, base stations and core networks almost bear all optimization tasks), the PHS standard does not specify the responsibility of base stations for network optimization, so some optimization of the wireless part can only start from the base station, such as clearing blind spots, reducing adjacent channel interference, and optimizing antenna elevation angles. Some functions can be handled by the mobile phone, such as predicting and analyzing network phase characteristics, initiating switching, changing the transmission power to improve the C/I ratio, and compensating for nonlinearities in transmission and reception.

Although the optimization of the wireless part is crucial and effective, it conflicts with the development strategy of Xiaolingtong whether it is started from the terminal or the base station. From the base station aspect, Xiaolingtong network requires a large number of base stations. According to power calculation and ignoring reflection factors, the number of PHS base stations is more than 25 times that of GSM base stations to cover the same area. Because of the lack of good planning in the early construction of Xiaolingtong network, the simple use of technologies such as increasing coverage, eliminating blind spots and multi-layer coverage, the optimization of base stations by operators is inevitably hindered by cost and capacity: the cost problem is reflected in the need to test, analyze, and update parameters of a large number of base stations, even moving base stations, which is complex and costly; the capacity problem is reflected in the fact that due to the disorder of base station planning and the statistical characteristics of PHS modulation methods, the conventional road test technology and analysis model used in GSM networks are far from meeting the needs of PHS. From the mobile phone aspect, Xiaolingtong's continuous cost reduction is due to technological progress, making the performance of new mobile phones no less than that of old mobile phones, but it also loses the opportunity to improve performance to achieve auxiliary network optimization.
Core network optimization is another way, but the optimization of the core network lies in balancing the access load to improve the success rate of access and switching. This requires knowing the base station coverage parameters, such as base station boundaries, overlapping areas, and cell instantaneous fading characteristics. These parameters are difficult to obtain. For example, PHS does not have clear base station boundaries and the fading is fast, resulting in unreliable drive test data. Therefore, it is impossible to rely solely on core network optimization.
The impact of dual-mode mobile phones on service quality

Wireless communication users can choose between networks and mobile phones. Taking PHS as an example, without considering the mobile phone factor, we study the interactive relationship between service quality (QoS), tariff/cost and users.
First of all, the number of users has a great relationship with QoS, and QoS is related to the cost of operating investment. The inflection point of N1(c) in Figure 1 is because due to technical limitations in the PHS network, excessive investment will reduce QoS. The intersection of the N1 and N2 curves represents the relatively stable number of users that the network has after the market development is relatively stable. All operators hope that this intersection will appear in the profit area.
The previous analysis curve about the network can be used for dual-mode mobile phones. Dual-mode mobile phones can support two networks at the same time. If we ignore the differences between the uU layer of GSM and the air interface of PHS and only consider the connection between the terminal and the core network, the result is equivalent to increasing the coverage of the wireless network. Its one-time access success rate (for two networks) is Pg+(1-Pg)Pp (Pg and Pp are the instantaneous connection success rates of each network). The handover success rate has also increased in a similar way. It can be said that dual-mode mobile phones have brought about an improvement in service quality. This has little impact on GSM, but has a huge impact on PHS, greatly expanding the roaming capabilities of PHS users. Without increasing operating costs, dual-mode users can freely compromise between rates and service quality, causing the intersection of the N1 and N2 curves to move to the right. As we will see below, dual-mode mobile phones using DSP processors can do more than this, and can even improve the service quality of the PHS network alone.

The structure and extension of dual-mode mobile phones

We understand the structure of dual-mode mobile phones according to the four-layer structure. From top to bottom, single-mode mobile phones are application layer (MMI), transport layer, network layer (L3), data link layer (L2) and physical layer (L1). In order to coordinate the two Uu interfaces and the physical layer, dual-mode mobile phones need to add a media convergence layer (MAC).

The earliest dual-mode mobile phone protocol stack architecture is relatively simple. It bundles two target modules on L1. PHS and GSM radio frequency and digital baseband are completed by different chips and processors respectively. In order to reduce costs, the baseband processing chip is generally ASIC, and the application layer above L2 including MMI is completed by a control processor, generally an ARM chip. Such an architecture bundles the wireless modems of P network and G network under the link layer, which is more collaborative than simply bundling two mobile phones. This architecture has several defects: 1. Although ASIC has low cost and power consumption, it does not have much advantage in cost and power consumption because it integrates two sets of wireless modems; 2. The digital baseband uses ASIC, which lacks flexibility, and this flexibility will play a huge role in optimizing performance in the solution of Figure 2b; 3. Similar to Wi-Fi, the two sets of RF and even antennas bring interference that cannot be ignored, and the circuit board wiring is also more complicated.

Another more advanced dual-mode solution. It uses only one RF chip to complete the demodulation of GSM and PHS, so the RF chip is required to be able to lock the frequency of PHS and GSM, and jump from one frequency to another in a short time. Its digital baseband is also completed by one chip, because the two decoding schemes are quite different, it is suitable to be completed by a DSP chip. In order to manage radio resources, arbitrate possible conflicts, and provide a unified interface to the upper layer, a media convergence layer (MAC) is embedded in L1, and both MAC and L1 above it are completed by MCU. Compared with solution a, the latter has the advantages of fewer hardware chips, small size and low power consumption, and because the digital baseband is completed by a DSP chip, it has better flexibility. However, this solution also has some defects: 1. In order to reduce cost and complexity, the RF chip uses zero intermediate frequency technology, which will bring or deepen the unfavorable factors such as local oscillator leakage, adjacent band interference and high-order intermodulation; 2. The cost and power consumption of DSP are greater than ASIC; 3. Shared wireless hardware resources will cause mutual exclusion conflicts.

In the future, dual-mode and even multi-mode mobile phones will go a step further and need to rely on network support. Similar to IP technology, it divides the network layer into two layers. The lower layer has routing capabilities and can establish heterogeneous network interconnection. After the user presets the strategy, the multi-mode mobile phone can dynamically jump between the GSM network and the PHS network during standby and call, and even use asymmetric wireless channels for uplink and downlink without affecting the user's operation. This technology solves the problem of network coverage in disguise, and the probability of disconnection is greatly reduced for both GSM and PHS communications.

PHS network optimization supported by DSP dual-mode mobile phones

We have previously had a general understanding of the architecture of DSP dual-mode mobile phones. In this section, we will focus on its features and significance for network optimization.
DSP dual-mode mobile phones use DSP chips to complete the work of the digital baseband part, including signal synchronization, waveform shaping, channel equalization, channel and source encoding and decoding, etc. In early GSM mobile phones, DSP chips were also used to improve reception performance, but they consumed a lot of power and were expensive. With the deepening of network optimization, the wireless MODEM with DSP core has been abandoned and ASIC or FPGA (in relatively high-level applications) has been adopted. The supply of ASIC is also monopolized by a few companies. The same is true for PHS. Dual-mode mobile phones give DSP architecture an opportunity, because different modulation modes and protocol standards require different baseband processing structures. If ASIC is used to complete it, the volume will increase, and wiring and power consumption are unfavorable factors. Today's DSP chips are very small in size and power consumption, and dual-mode mobile phones can be regarded as higher-end requirements, and the increased cost due to DSP can be ignored. In terms of chip supply, TI's OMAP and ADSP's Hermes both adopt the DSP+ARM architecture, which can be regarded as tailor-made for multi-mode smartphones in terms of cost, power consumption and size.

DSP dual-mode mobile phones can do more than ASIC. Because the network optimization of GSM in China is already very complete, the application of DSP has no substantial effect on GSM. However, PHS is different. As mentioned earlier, PHS network optimization is very difficult, which can be summarized as follows: 1. There is no clear boundary between base stations; 2. There is a lack of frequency planning; 3. There is a relatively fast fading. The first problem leads to disordered switching, the second problem causes co-channel and adjacent channel interference, which in turn leads to passive switching channels or base stations, and the third problem causes the critical user capacity of the cell to change rapidly. Another problem is that there is no reasonable strategy for PHS switching. In the past, neither the terminal nor the network could analyze the channel characteristics in real time. In addition, because the jitter of the PHS channel is much higher than that of GSM, the results of the road test instrument are unreliable. This is often manifested in urban areas with complex terrain, such as places with many high-rise buildings. After the service quality reaches a certain level, it no longer increases with the increase in the number of base stations, and may even deteriorate.

The PHS mobile phone with DSP architecture is far superior to the mobile phone with ASIC core in terms of transceiver performance. DSP can complete the analysis of channel parameters, select the corresponding pre-stored processing program according to the different statistical characteristics of the channel, and the interaction with the core network can analyze the location of the terminal and predict the reliability probability of the base station. DSP can also perform nonlinear compensation for transmission, which is not possible for ASIC. The enhancement of transceiver performance increases the instantaneous user capacity, and the analysis of the channel provides a more reliable basis for switching. Let's take the area with multiple coverage of large and small base stations as an example. If we simply use power factor to select a base station, it may be selected because the small base station is larger at the moment of measurement, and the fluctuation of the small base station is faster, which has a great harm to the call quality of roaming users, especially in urban areas, where the number of users is near the critical point, which has fatal damage to the entire network. Real-time measurement of channel parameters can actively initiate switching and speed up switching, which is conducive to the optimization of load distribution in the cell.