Research Article | Open Access
Coordinated Semidirectional Distributed Antenna System with Capacity and Energy Efficiency Analyses for Cloud Cellular Network
In order to further explore the coordination gain from the cloud cellular network, the Coordinated Semidirectional Distributed Antenna System (CS-DAS) is proposed with deploying 180-degree sector antennas. Based on the CS-DAS, the coordinated cellular deployment structure is established with corresponding communication procedure designs. The handover process for CS-DAS is designed with less handover rate and signaling overheads. The system performances of ergodic capacity and Energy Efficiency are derived, with a strong focus on the downlink performances, and analyzed with comparisons of current cellular structure. The numerical expressions of ergodic capacity are obtained with improvements of coordinated gain. The Energy Efficiency with different transmission modes is also investigated. The simulation results verify the performance improvements.
With increasing demands for versatile mobile data services, the mobile communication systems are triggered to further evolve toward much higher data rates and user consistent experiences on both the cell center and cell edge. Meanwhile, since the Information and Communication Technology (ICT) is playing an increasing role in global greenhouse gas emissions, higher Energy Efficiency (EE) should also be focused on along with the capacity booming requirements for future mobile communication systems .
In order to fulfill the above requirements, Coordinated Multipoint (CoMP) transmission/reception technique was implemented in Long Term Evolution-Advanced (LTE-A) system as a typical application of coordinated communication techniques. CoMP is proved to provide higher data rates and better cell edge performances [2, 3]. With the aid of CoMP, the EE performance could also be improved by the coordinated transmission and reception with utilizing diversity gains.
Furthermore, the Remote Radio Head (RRH) was also introduced into the CoMP with more flexible cellular deployments, which can fully explore the capacity improvement from coordination transmissions . The antennas could be deployed remotely from the Base Stations (BSs) with only handling the wireless signal transmission/reception, while the baseband unit of BSs could focus on the centralized signal processing, resource managing and scheduling, and so forth with the cloud computing techniques. Then, the cloud cellular network concepts are proposed, among which the Cloud-infrastructure Radio Access Network (C-RAN) architecture is the representative approach . C-RAN aggregates the resources in the RAN through the centralized management manner with the Virtual BS Pool. The Radio Frequency fronts are deployed with connections to the Virtual BS Pool in a distributed way. The centralized resource allocation and scheduling for the users are supported, which makes the optimization of wireless networking strategies easier. There are years of research and trial network tests with the C-RAN, by which the advantages are already verified.
The distributed remote antennas in the C-RAN provided the possibilities for further exploring the coordination gain, but the challenges of how to deploy the distributed remote antenna for maximizing the system capacity and EE also appeared. Therefore, the cellular deployment structure for coordinated transmission in the cloud cellular networks should be further researched.
The basic distributed and coordinated topology for the cellular network should be the Distributed Antenna Systems (DAS) proposed in 1987 with the fixed cellular structures . The distributed cellular deployments have been evolved in decades with Group Cell Architecture , Distributed Wireless Communication System (DWCS) , and also CoMP system. Both the omnidirectional antenna and the sector antenna were researched within the distributed network deployments for obtaining better capacity performances.
These studies have identified potential advantages of DAS such as reduced transmit power, lower handover ratio, and increased system capacity. But the design for the coordinated cellular network structure is the key issue for the distributed network deployment, which also impacts the handover process in the coordinated distributed networks. Currently, 120-degree sector antennas, that is, antennas with a horizontal coverage of 120°, are employed in the system for higher frequency reuse and system capacity. But the 120-degree sector antenna also causes more sector-edge effects. However, there is little concern on the deployment of other types of sector antennas in the mobile communication networks.
Actually, the 180-degree sector antenna has been utilized in radar and satellite communication systems (more information could be found in [8–10]). For the real hardware products, the antenna gain of 180-degree sector antenna is around 15 dBi . With the introduction of 180-degree sector antenna into the cloud based cellular networks, the horizontal coverage features of 180-degree sector antenna will contribute to the decrease of cross-coverage in the cell edge. The handover process could also get benefits from that. Moreover, the coordinated gain brought by the distributed deployment of 180-degree sector antenna will be further improved, also with system EE performances.
Therefore, we proposed the Coordinated Semidirectional Distributed Antenna System (CS-DAS) with 180-degree sector antennas, which will establish better coordinated environments. The main contributions of this paper are as follows:(i)The CS-DAS is designed for further exploring the coordinated gain in cloud cellular networks with centralized processing.(ii)The handover procedures in the CS-DAS are provided with better support of user mobility in coordinated networks, which could alleviate frequent handovers and decrease the signaling overhead during the handover process.(iii)The ergodic capacity for downlink scenarios of the CS-DAS is improved further with the coordinated gain in the distributed deployments.(iv)The system Energy Efficiency with different transmission modes for the downlink CS-DAS is derived.
The rest of this paper is organized as follows. The system model with the typical deployment scenario and features of the CS-DAS are described in the next section. In Section 3, the ergodic capacity and EE performances are provided with the statistical expressions derivation. Section 4 presents the performance evaluation with simulation results. Finally, there comes the conclusion.
2. System Model
The typical distributed cellular deployment scenario with CS-DAS is shown in Figure 1. Based on the designed CS-DAS architecture, each cell consisted of one Base Station (BS) at the cell center, which is separated with one Central Antenna Unit (CAU) and six Remote Antenna Units (RAUs) located at the cell edge. The signal processing functions are implemented by the BS with a centralized manner, while the antenna units (AUs) mainly charge the wireless signal transmission/reception via dedicated bidirectional backhauls (e.g., optical fibers) with the BS. This is a small scale cloud cellular structure. If all of the BSs are connected to one Virtual BS Pool, there becomes the typical C-RAN deployment scenario.
The CAU is located at the cell center and deployed with the omnidirectional antenna, while the RAUs are located at the boundary to the adjacent cells and deployed with 180-degree semidirectional sector antennas which are only able to transmit and receive signal at 180-degree horizontal coverage towards the adjacent cells.
For example, in Figure 1 has the CAU named and six RAUs named . The RAU is located at the boundaries of and with the semidirectional antenna towards the center of , which creates the cross-coverage area for coordinated transmissions inside of the coverage of CAU in . Meanwhile, the RAU , which belongs to , is also located at the boundaries of and with the semidirectional antenna towards the center of . There will exist cross-coverage areas, which are potential for coordinated transmissions. The actual covering region for the BS in the CS-DAS is expanded and the cell edge will be overlapped with two or three AUs which belong to different BSs.
The symbols and of the RAU represent the relations of the RAU with its belonging BS and location. The symbol denotes that the RAU belongs to and the symbol means the coverage region of is located inside .
In order to get better understanding of the CS-DAS cellular deployment structure, Figure 2 is plotted with the examples that support the coordinated communications for users. Three BSs are involved in the coordinated communications with the User Equipment (UE). As shown in Figure 2, is locating at the cross-coverage region of the CAU of and the RAU of . The coordinated communication can be established between the two BSs. While is locating at the cross-coverage region of the CAU of , the RAU of , and also the RAU of , the CoMP with three BSs will be set up for . The cell edge performance will be further improved.
The resources coordination for the UE will be handled by the centralized processing BSs via CoMP techniques. The RAU’s resource usage will have constraints from the adjacent cells at which the RAU actually locates to avoid the interference. The current intercell interference coordination solutions, such as Fractional Frequency Reuse (FFR) and Soft Frequency Reuse (SFR) implemented in LTE networks, can still be applied with modifications for the CS-DAS architecture.
As shown in Figure 1, because all the RAUs are distributed within the coverage areas in adjacent cells, the frequency allocation for the reusing policy is different with current FFR or SFR schemes. For example, the RAU is located at the boundaries of and with the semidirectional antenna towards the center of , which creates the cross-coverage area for coordinated transmissions inside of the coverage of CAU in . So, the frequency allocated to the RAU for reusing should be from the available frequency set of rather than the available frequency set of its connected . Furthermore, this frequency reuse method could also guarantee the coordinated communication between RAU and CAU , because they are reusing the same frequency set of . The frequency reuse policy for the CS-DAS structure is worth to be further researched in the future, which could bring more benefits to this architecture.
Based on the deployment rules, the CS-DAS architecture can bring more benefits besides the coordinated transmissions. With more and more antennas deployed inside of the cellular network, the handover will be more frequent, which will degrade both the user experiences and the network performances. But thanks to the coordinated coverage areas created by the CS-DAS, there will be less handovers and the moving users’ experiences will be improved, while the network performances will also be guaranteed with less signaling overheads associating the handovers.
In order to show the benefits brought by the CS-DAS, a further perspective is given with one application scenario as shown in Figure 3, where the UE performs handover through two adjacent cells ( and ) and the communication/handover process will be divided into three procedures according to the CS-DAS structure.
Procedure 1. The UE moves from the cell center to the cell edge of , where the cell edge or cell center can be simply determined by the user received Reference Signal Received Power (RSRP). During the UE’s moving inside of , the UE maintains continuous communication with from the CAU . When the UE moves into the cell edge of , the UE will additionally get access to with the RAU acting as the CoMP transmission point. When UE’s access to is accomplished, the UE will get service from both and with two AUs’ coordinated transmission/reception to get performance gain from CoMP.
Procedure 2. The UE moves from the cell edge of to the cell edge of . The serving AU set will consist of from and from instead of the former serving AU set in Procedure 1 with the CAU and the RAU . Please note that the current serving cells are still and , which means that the handover only occurs inside of the coordinated set. The handovers will be easily handled without any extra overheads.
Procedure 3. The UE moves from the cell edge to the cell center of . While the UE moves out of cross-coverage region of the RAU of , the coordinated transmission/reception will be terminated and the serving AU set will be updated to only from . The handover will be finished with withdrawing from the serving cells. is always involved in the whole handover process from Procedure 2. The UE handover performance will be guaranteed and the signaling exchanges will be less than current handover procedures in LTE systems.
From the procedures given above in the CS-DAS, it can be concluded that when a UE moves from one cell to another, the UE can always stay in communication with at least one BS. The handover processes from to achieve the seamless handover function. will guarantee the UE’s performance until it moves into the cell center of . And has already taken parts in the responsibility of serving the UE even from where it still stays in the cell edge of . The whole procedure just seems like the relay race.
Meanwhile, as analyzed in , the rate of handover mainly depends on the relative signal strength differences from adjacent antennas. Based on the CS-DAS, the signal strength of desired RAU is relatively bigger than other undesired CAUs or RAUs at the cell edge (low path-loss region of the desired RAU). The total number of handovers can be decreased. What is more, from the features of 180-degree semidirectional sector antennas, such as the ability of mitigating Doppler Shift , CS-DAS can also be used on highway deployment scenarios and achieve more improvements.
3. Ergodic Capacity Analyses for CS-DAS
For the performance evaluation of CS-DAS, the ergodic capacity is one of the most important metrics for cellular deployment structures. In this section, the ergodic capacity for CS-DAS structure will be studied. We mainly focus on the downlink performance in this paper.
The CS-DAS cellular deployment structure used for capacity analysis and evaluation is shown in Figure 2, where gets access to through the CAU and through the RAU . The serving AU set consists of . gets access to the serving AU set .
Assume that there are cells ( AUs) in the research region. The frequency resource used by is assigned as . As may be assigned to the cell center user or the cell edge user with coordinated AUs (CoAU-, or 3) in , there are definitely one part of signal power on from the CAU and no more than two parts of signal power on from the RAUs .
Generally, we can deduce the common cases that there will be one part of signal power from the CAU and no more than two parts of signal power from the RAUs of adjacent cells. Thus, there are totally no more than parts of signal power on for the users in the research area.
The discrete-time received signal formula for the analyzed UE is given aswhere is an indicator factor that represents whether there exists signal power on from the AU or not (i.e., or 1) and . denotes the long term shadow fading modeled by log-normally distributed random variable with standard deviation and mean value , with both being in dB. The mean value is defined as , with the path-loss exponent , constant , and distance between the AU and the UE. denotes the gain of the frequency flat Rayleigh fading channel, modeled by independent and identically distributed complex Gaussian variable with unit variance . represents the signal transmitted from the AU and is the additive noise.
Let and denote the sets of indices representing the coordinated AUs and other remaining AUs. Equation (1) can be rewritten as
The UE received signal power iswhere is the average transmit signal power of the AU , represents the desired part of signal power, and denotes the undesirable part of signal power (i.e., the interference).
For simplicity of the capacity analyses, we mainly consider the path-loss effects during the ergodic capacity analysis for both the CS-DAS and traditional cellular network. Then, the relationship between Signal to Interference plus Noise Ratio (SINR) and planar position can be expressed as one-to-one map and the ergodic capacity can be calculated as where is the coverage area of one cell in the CS-DAS, is the Probability Density Function (PDF) of UE located at position , and is the function of SINR. Because the coordinated communication mode is embedded in CS-DAS architecture, we cannot obtain the closed-form capacity with interference considered. Therefore, we choose the integral capacity expressions as (4) for conducting the evaluations on capacity and EE performance improvements brought by the CS-DAS architecture.
4. Energy Efficiency Analyses for CS-DAS
According to the coordinated communications implemented in the CS-DAS, the downlink Energy Efficiency performances of the cell edge users served with the coordinated AU set will be the focus. Due to many previous research works, the EE performance of users at the cell center served only by one CAU scenario will not be of concern in this part.
During the following analyses, we assume the system is interference-limited; thus the effect of noise is negligible. The instantaneous SIR of UE with CoAU- coordinated transmission mode can be written as
EE is defined as the ratio of achieved transmission rate over the total power consumption, which is given aswhere is the transmit power of the coordinated AU and is the circuit power consumption which has been discussed in . For simplicity, the circuit power consumption is defined as a constant (given in ) in the following analyses.
The Cumulative Distributed Function (CDF) of EE can be obtained aswhere denotes the received signal vectors, , and represents the conditional CDF of EE conditioned on . And the joint PDF of can be obtained in 
The EE performances of different scenarios as two AUs and three AUs involved in the coordinated transmission will be analyzed separately.
Scenario 1 (CoAU-2, coordinated transmission with 2 AUs). As shown in Figure 2, is in CoAU-2 transmission mode. For simplicity of analyses, only the main interference will be considered during the following analyses. The interference for mainly comes from the adjacent cells, which are the CAUs of and , also with the RAUs inside of due to their beam directions. Thus, the SIR for can be written aswhere is Boolean and with the coordinated AU set establishment rules described in Section 2. represents the other remaining interference modeled by its expected value and .
If , the SIR in (9) will be maximum and is given as
Considering the form of , can be the substitution of and can be obtained in the actual networks. For given values of log-normal shadowing variable , follows the Chi-squared distribution with 2 degrees of freedom. And follows the exponential distribution conditioned on and the PDF is , where .
The numerator part of (10) follows the weighted Chi-squared distribution conditioned on and , based on the sum principle of two independent random variables ; the corresponding PDF can be derived aswhere .
Similarly, the denominator of (11) is and its conditional PDF is given aswhere .
According to the transformation rules given in , the conditional PDF will be obtained. And the conditional CDF is given aswhere , , and .
Similarly, by defining , . The conditional CDF can be obtained aswhere and .
The EE for CoAU-2 () is given as ; thus
Scenario 2 (CoAU-3, coordinated transmission with 3 AUs). As shown in Figure 2, is in CoAU-3 transmission mode. Similarly, the SIR for can be written aswhere , , and .
Similar to the analyses of scenario CoAU-2, by defining , , and , the conditional CDF can be derived aswhere , , , and .
The EE for CoAU-3 () is given as ; thus
For comparisons, based on the traditional cellular structure without coordinated transmission as the baseline, the users will get service from only one CAU. The ergodic capacity of the traditional cellular structure still follows the same formula form as (4), but the interference from the RAUs will not exist. The SIR for and can be expressed in the unified form and written as . The conditional CDF of EE can be derived aswhere and . for and can be derived with formulas (7), (8), and (19).
5. Performance Evaluations
In this section, the performance evaluation with simulation results for both the downlink ergodic capacity and EE is conducted, and the comparison with traditional cellular structure will be introduced.
Simulations are operated for the CS-DAS and traditional cellular structure under the same conditions. A general 3-layer cellular structure with 19 BSs is considered in the simulation. The simulation parameters and settings are listed in Table 1 based on 3GPP evaluation model . Because we mainly focus on the sector antenna’s horizontal coverage capabilities for multicell coordination, the vertical coverage features are not involved in the simulation. The relationship between the average transmit power of CAU () and RAU () is shown as with the consideration to make the same attenuation with the path loss at the coverage boundary.
Because there are two coordination modes above for the coordinated communications, in order to analyze the EE of the two modes of two AU coordinated transmissions and three AU coordinated transmissions, we predetermine and locations to guarantee the desired coordinated communication mode.
5.1. Simulation Results for Ergodic Capacity
In order to compare the capacity performances of CS-DAS with traditional cellular structure, we define the capacity gain as the capacity difference to the traditional cellular structure capacity ratio, given aswhere is the capacity of CS-DAS and is the capacity of traditional cellular architecture. and can be obtained from formula (4) in Section 3.
The influences of intersite distance (ISD) and the average transmit power of CAU are both considered in the ergodic capacity simulation, and the result is shown in Figure 4. It can be found that the least gain of CS-DAS is about 22.08% and the difference between maximum and minimum gains is less than 0.26% (the maximum is 22.34%) which means that CS-DAS can almost bring the same capacity gain in different deployment scenarios with different ISDs or transmission powers.
In order to show the performance especially in the cell edge with coordinated gain, the CDF of the ergodic capacity for CS-DAS and traditional cellular structure are evaluated and the results are shown in Figure 5. The whole throughput performance of CS-DAS is better than the traditional cellular structure as shown in Figure 4. The cell edge throughput performance (5% of the CDF) is further better with the CS-DAS.
5.2. Simulation Results for Energy Efficiency
During this subsection, simulations of the EE are performed for and with different coordination transmission scenarios introduced in Section 4. The simulation is carried out by taking both path loss and Rayleigh fading with the same parameter setting as Table 1.
Scenario 1 (CoAU-2). The simulation is performed on with CoAU-2 coordinated transmission scenario. And the traditional cellular structure transmission with only one CAU serving is also simulated for comparison. For CoAU-2 coordinated transmission scenario, as shown in Figure 2, the coverage direction of the RAU is opposite to ; is established. For the situation that the interfering signal power from is just one part (i.e., ), the EE is indexed as ; represents the number of RAUs from which the interference comes. Meanwhile, on the situation that the interfering signals from the RAUs of have two parts (i.e., ), which means the same resource is allocated for 3-AU coordination transmission mode with two adjacent RAUs and one CAU (the numbers of the adjacent RAUs should be and ), the EE for different interference sources are indexed as when and .
Derived from the above expression, we can get that and ; the EE for can totally be expressed as 10 results with different interference resources (including , where ).
The CDFs of the EE , , , and are obtained by using the derived equations for the CoAU-2 scenario and the traditional cellular structure in Section 4. The results are plotted in Figure 6. The more right position the curve lies in, the better performance it indicates. Demonstrated by the results, the EE performances of CoAU-2 scenario are much better than the traditional cellular structure and the best situation of CoAU-2 () produces the most improved EE performance with . The sorted right-left order of and is given as , , , , , , , , and . Actually, the lower the interference suffered from the interfering RAU, the better EE performance will be gotten.
Furthermore, taking a close eye on the curve of , the sorted right-to-left order is given as , , , , and , while the distances between the RAU and are shown as , which means the farther from to the interfering RAUs, the better EE performance. So, according to this factor, we can further design the resource allocation and intercell interference coordination strategies for better EE performances.
Scenario 2 (CoAU-3). The evaluation for the CoAU-3 scenario is conducted on comparing with the traditional cellular structure. The basic deployment is also shown in Figure 2; the coverage directions of the RAUs , , and are opposite to ; thus , , and . The interference signals from and are considered in the simulation, and there may be 0, 1, or 2 parts of interference signals from the above BSs. Totally, there will be 80 possibilities for EE of . To simplify the simulation results and without loss of the generality, we will fix the RAUs of and as two main interference sources. By defining for the EE with parts of interference power from and parts of interference from , the simulation result is shown in Figure 7. It can also be concluded that the CoAU-3 scenario is better than the traditional cellular structure in terms of EE and shows the best performance.
This paper proposed the Coordinated Semidirectional Distributed Antenna System to further explore the coordinated gain with deploying 180-degree sector antennas in the downlink scenario of cloud cellular networks. With special designed topology of antenna deployments and coordinated antenna set establishment, the new designed handover procedure has advantages of decreasing handover rate and signaling overheads in the coordinated communications. Furthermore, we can get both the capacity gain and Energy Efficiency performance improvements. The ergodic capacity and the Energy Efficiency in the downlink are deduced and analyzed with the numerical expressions. Moreover, the EE performances of given users for different coordinated scenarios with CAU and RAUs are derived. The simulation results show that the downlink capacity gain brought by the CS-DAS can be at least 22.08% over the traditional cellular structure. The downlink EE performance improvements in CS-DAS are also identified with different coordinated scenarios by simulations. The CS-DAS gives a new deployment trial for exploring the coordinated gain in centralized cloud cellular networks.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
This paper is supported by the Nature and Science Foundation of China under Grants nos. 61471068 and 61421061, Beijing Nova Programme no. Z131101000413030, and Collaborative Project 2015DFT10160.
- C. Han, T. Harrold, S. Armour et al., “Green radio: radio techniques to enable energy-efficient wireless networks,” IEEE Communications Magazine, vol. 49, no. 6, pp. 46–54, 2011.
- 3GPP, “TS 36.814, Further Advancements for E-UTRA Physical Layer Aspects, v9.0.0,” 2010.
- A. Nosratinia, T. E. Hunter, and A. Hedayat, “Cooperative communication in wireless networks,” IEEE Communications Magazine, vol. 42, no. 10, pp. 74–80, 2004.
- China Mobile Research Institute, “C-RAN, the road towards green RAN, version 3.0,” White Paper, 2013.
- A. A. M. Saleh, A. J. Rustako Jr., and R. S. Roman, “Distributed antennas for indoor radio communications,” IEEE Transactions on Communications, vol. 35, no. 12, pp. 1245–1251, 1988.
- X. Tao, X. Xu, and Q. Cui, “An overview of cooperative communications,” IEEE Communications Magazine, vol. 50, no. 6, pp. 65–71, 2012.
- S. Zhou, M. Zhao, X. Xu, J. Wang, and Y. Yao, “Distributed wireless communication system: a new architecture for future public wireless access,” IEEE Communications Magazine, vol. 41, no. 3, pp. 108–113, 2003.
- D. H. Pham and T. Wada, “Application of array antenna for a high-speed ISDB-T reception,” in Proceedings of the 2nd International Conference on Communications and Electronics (ICCE '08), pp. 351–354, IEEE, Hoi An, Vietnam, June 2008.
- Y. Wang, Y. Sun, and H. Yang, “Analysis and design of a satellite-borne wide-beam conical quadrifilar helical antenna,” in Proceedings of the International Symposium on Signals Systems and Electronics (ISSSE '10), vol. 1, pp. 1–3, Nanjing, China, September 2010.
- D. Lopez-Perez, I. Guvenc, and X. Chu, “Mobility management challenges in 3GPP heterogeneous networks,” IEEE Communications Magazine, vol. 50, no. 12, pp. 70–78, 2012.
- M. D. Austin and G. L. Stueber, “Exact cochannel interference analysis for log-normal shadowed Rician fading channels,” Electronics Letters, vol. 30, no. 10, pp. 748–749, 1994.
- X. Xu, X. Tao, C. Wu, Y. Wang, and P. Zhang, “Capacity analysis of multi-user diversity in generalized distributed cellular architecture,” in Proceedings of the 6th International Conference on ITS Telecommunications Proceedings, pp. 533–536, IEEE, Chengdu, China, June 2006.
- S. Cui, A. J. Goldsmith, and A. Bahai, “Energy-constrained modulation optimization,” IEEE Transactions on Wireless Communications, vol. 4, no. 5, pp. 2349–2360, 2005.
- C. He, B. Sheng, P. Zhu, and X. You, “Energy efficiency and spectral efficiency tradeoff in downlink distributed antenna systems,” IEEE Wireless Communications Letters, vol. 1, no. 3, pp. 153–156, 2012.
- W. Roh and A. Paulraj, “Outage performance of the distributed antenna systems in a composite fading channel,” in Proceedings of the IEEE 56th Vehicular Technology Conference (VTC '02), pp. 1520–1524, Vancouver, Canada, 2002.
- A. B. Gershman, N. D. Sidiropoulos, and A. B. Gershman, Space-Time Processing for MIMO Communications, John Wiley & Sons, New York, NY, USA, 2005.
- C. M. Grinstead and J. L. Snell, Introduction to Probability, American Mathematical Society, 2nd edition, 1997.
Copyright © 2016 Xiaodong Xu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.