Corresponding author: Z. Zhu, College of Electronics and Information Engineering, South-Central University for Nationalities, Wuhan, Hubei 430074, China. (firstname.lastname@example.org)
 Detecting Doppler frequency shifts from ionospheric high-frequency echoes is an important way to study ionospheric disturbances. This paper presents and realizes a new mode for observation of ionospheric disturbances using a combination of coded pulses and echo phase measurement analysis in ionospheric vertical sounding based on the Canadian Advanced Digital Ionosonde (CADI) platform. Experimental results show that the newly developed mode for observation of ionospheric disturbances on CADI can acquire accurate Doppler ionogram (Dopplionogram) and obtain temporal and spatial variations of the velocity of ionospheric disturbances in real time so that it has essential value in observation and research of ionospheric disturbances. The application of the new mode for observation of ionospheric disturbances in ionospheric vertical sounding opens up a new, effective way by which much more ionospheric information can be acquired with existing common ionospheric sounding instruments.
 Ionospheric vertical sounding method invented in the 1920s showed the existence of the ionosphere and ionosonde instruments designed to use this method have been widely used for ionospheric observation and research [Hunsucker, 1991; Davies, 1989]. With over 70 years of development the ionosonde went through such several phases from manual operation to automatic operation, from analog form to digital form, from single virtual height measurement to multiple parameter measurements like amplitude, phase, polarization, Doppler frequency shift, arrival angle and so on simultaneously. In the middle of the 1990s with the rapid development of Internet communication and hardware and software technology of personal computer, modern digital ionosondes with lower power consumption and networking function were developed and put into use [Reinisch, 1996; Wright and Pitteway, 1999; Ning et al., 2000]. Nowadays, a digital ionosonde is a universal instrument for observing the ionosphere from the ground.
 Various digital ionosondes presently in use can obtain the curve of variation of the echo's virtual height with the scan frequency, scale various ionospheric characteristics parameters automatically or semiautomatically, and also invert the ionogram and obtain the electron density profile below the ionospheric F2 layer peak [Zhu et al., 2007a; Ding et al., 2007]. In addition, these instruments can also usually transmit several pulses and do FFT analysis of these pulses. As a result, the Doppler frequency shift of these high-frequency signals can be obtained. However, owing to the limited duration time of the transmitted pulses, the precision of the Dopplionogram obtained by the above method is poor and cannot entirely meet the demand for observation and research of ionospheric disturbances. Although the drift mode for ionospheric observation used in ionospheric vertical sounding can get the Doppler frequency shift with higher precision, the number of the observed frequencies is few. Consequently, the height profile of the ionospheric disturbance cannot be fully obtained [Bibl and Reinisch, 1978; Wan et al., 1993].
 In order to improve the measurement precision of Doppler frequency shift and other echo parameters, Wright and Pitteway  proposed a new dynamic ionosonde named Dynasonde. In the Dynasonde, the combined pulses are elaborately designed, the receiving antenna array are skillfully disposed, and the echoes' phases are accurately acquired. With these phases, Doppler frequency shift information and polarization information of the echo can be calculated [Tsai et al., 1993]. Yuan et al. selected 16 channel data in frequency domain recorded by Digisonde ionosonde in the ionogram mode, and transformed these frequency domain signals into time domain signals using inverse FFT algorithm, and computed the phase change of these time domain signals with time using a linear regression of least squares method, and finally acquired the accurate Dopplionogram. However, because of the limitation of Digisonde in hardware design, this method can only be done in an off-line mode.
 For real-time observation and analysis of the ionospheric disturbances, we imported a Canadian Advanced Digital Ionosonde (CADI) [Gao and MacDougall, 1991; Huang and MacDougall, 2005], and developed a set of new application programs for system control, data analysis, and data processing under Windows operation system instead of the original programs that ran under DOS operation system, and we realized a new mode for observation of ionospheric disturbances, in which we acquire accurate Dopplionograms for observation and analysis of ionospheric disturbances in real time. In this study, the hardware platform for developing the new mode for observation of ionospheric disturbances is described in section 2. Then the real-time acquisition method of Dopplionograms with high precision is presented insection 3. The experimental results and the related discussions are given in section 4. Finally we make conclusions in section 5.
2. Hardware Platform for the New Mode for Observation of Ionospheric Disturbances
 In order to realize the new mode for observation of ionospheric disturbances in vertical sounding, a digital ionosonde with open structure is needed. On this ionosonde platform, the combined pulses should be designed and raw data of the echo should be processed in real time so that continual accurate Dopplionograms can be acquired in a time interval of 2–5 min. For this study, the CADI digital ionosonde can satisfy the above demand, therefore we selected CADI as our experimental platform. CADI is a low-cost, full-featured, flexible modern ionosonde ideal for both routine ionospheric monitoring and scientific research, which integrates pulse compression technology and modern advanced microelectronics application technology [Grant et al., 1995; MacDougall and Li, 2001; Fagundes et al., 2005; Zhu et al., 2007b; Zhao et al., 2009]. The system functional block diagram of CADI is shown in Figure 1. CADI uses a personal computer as system main control and processing unit. The PC controls a Direct Digital Synthesis (DDS) chip via the ISA bus to generate a series of encoded scan frequency pulses of radio wave signals, which are vertically transmitted upward to the ionosphere. The echoes reflected from the ionosphere are received by the receiver of CADI and are down converted into intermediate frequency signal, which is demodulated in phase into the two orthogonal signals, namely I signal and Q signal. After the IQ signals are sampled, quantized and decoded, I and Qsignal are transported to PC via the ISA bus. After data collecting and information processing, the curve of variation of virtual height with scan frequency, that is ionogram, can be acquired in real time, meanwhile the raw data of the echoes can also be saved to the PC for off-line processings.
 For the new mode for observation of ionospheric disturbances, CADI also possesses other characteristics as described in the following sections.
2.1. DDS Technology Convenient for Improving Frequency Transient Change Performance
 DDS (Direct Digital Synthesizer) technology, which is programmable, operation flexible and user friendly convenient, has been developed accompanied with the development of microelectronics technology and VLSI (Very Large Scale Integration) technology. With features of short conversion time and high precision of frequency signal, DDS is widely used in various digital frequency source systems. A DDS system is composed of accumulator, adder, waveform ROM, D/A converter and low-pass filter.
 In a CADI system, DDS chip Q2334 is interfaced with MCU MC68HC811 and is controlled by the MCU to generate the needed frequency signals. With the clock signal frequency of 50 MHz, Q2334 can generate various frequency signals from 1 MHz to 20 MHz with frequency resolution 0.00465 Hz and frequency conversion time 0.62 μs. This provides favorable conditions for designing combined pulses for the new mode for observation of ionospheric disturbances. In addition, Q2334 has two channels and can simultaneously generate two frequency signals, and this satisfies our demand for frequency signals, one radio frequency signal for transmitting, another local oscillator signal for down converting.
2.2. Open Software Platform of CADI Convenient for System Redevelopment
 The CADI digital ionosonde controls and operates the whole system by DOS software installed in PC, and the hardware port is specific and the flowchart of software is clear. Therefore CADI has good condition for system redevelopment. Compared with DOS operation system, Windows has some outstanding technology characteristics, for example, standard user interface, supporting multiple tasks, excellent memory management and independent peripherals and so on. In addition, Windows has very powerful networking function, graphic display function and multimedia function. In the new mode, we select Windows 2000 as the operation system, which has also greater improvement in system security, stability and reliability compared with older editions of Windows.
 On the CADI hardware platform and under open system resources, we designed and realized new application programs for system control, data acquisition, data analysis, data processing, and running monitoring under the Windows 2000 operation system.
 Because CADI controls each function module via ISA bus, I/O port reading and writing must be done in Windows application programs. This means that we need to write driver programs for application programs under Windows 2000 operation system. In this study, we write Windows driver programs by using WDM (Windows Driver Model) and integrate it in system by DLL (Dynamic Link Library) format.
3. Acquisition of Accurate Dopplionogram
 Detecting Doppler frequency shifts from ionospheric high-frequency echoes is an important way to study ionospheric disturbances and it can detect the velocity of ionospheric reflection surface so as to further study the movement and variation of the ionosphere [Li, 1983; Zhang and Tschu, 1988].
 In order to carry out observation of ionospheric disturbances in ionospheric vertical sounding, we design and develop a new mode for observation of ionospheric disturbances on a digital ionosonde platform, by which continual accurate Dopplionograms with a time interval of 2–5 min are acquired in real time and the curve of variation of ionospheric disturbances with time and scan frequency is analyzed and displayed. In this new mode, Doppler frequency shift is acquired by means of measuring the variations of phase of the time domain signals with time, instead of the former method in which Doppler frequency shift is acquired by calculating FFT of time domain signals. However, it must be pointed out that this measurement method demands narrowband signals with appropriate signal-to-noise ratio (SNR). The actual measurement results reveal that the echo signals received by CADI have proper SNR, thus they can satisfy the above demand for ionospheric vertical sounding. Moreover, while the new mode for observation of ionospheric disturbances works, the routine digital ionogram mode is not interrupted and can work simultaneously.
 The phase of high-frequency signals arriving at the receiver of the CADI digital ionosonde is expressed inequation (1). The Doppler frequency shift of the echoes reflected by the ionosphere is described in equation (2). From equations (1) and (2), equation (3)can be deduced to calculate Doppler frequency shift of high-frequency signals
where Φ denotes the phase of the received high-frequency signals; Φ0 denotes the original phase; ω0 denotes the frequency of the transmitted signals; c is the velocity of the electromagnetic wave; P denotes the phase path from the transmitting spot to the receiving spot; s is the propagation path of the signals; μdenotes the phase refractive indices of the high-frequency signals; and Δωdenotes Doppler frequency shift of the high-frequency signals.
 In the new mode for observation of ionospheric disturbances, the scan frequency manner of the transmitted pulses is shown in Figure 2, where PRF denotes the repeat frequency of the transmitted pulses, IPP denotes the time interval of two adjacent pulses and each short horizontal line denotes the number N of transmitted pulses at each sounding frequency. Figure 3 shows the time sequence of the transmitted pulses on CADI. When a pulse is transmitted at a certain sounding frequency, each of two receiving channels collects M pairs complex signals data corresponding to different heights and all the 2 × M × N pairs complex data are stored in memory of PC. These data from the two channels are combined to separate the ordinary wave (O Wave) from the extraordinary wave (X Wave) by a software method. Therefore narrowband echo signal is acquired. Then these narrowband echo signals are used to calculate phase value and the ambiguity of these phases is unwrapped. Finally the Doppler frequency shift Δω is calculated by means of the least mean squares linear regression (LMSLR) algorithm. The schematic diagram of unwrapping the phase ambiguity and the least mean squares linear regression algorithm is shown in Figure 4. After all the sounding frequencies are processed with the above method, Dopplionogram can be acquired.
 The method of acquiring Doppler frequency shift by echo phase measurement, however requires the condition that both SNR and the number N of the transmitted pulses at the same frequency cannot be too low. Otherwise, the measurement error of Doppler frequency shift will be large. In order to estimate what SNR and the selected N are suitable for the method to work well, some experiments of computer simulation were done for this. Suppose that there was a monochromatic signal mixed with Gauss noise expressed in equation (4):
where ej2πft denotes the monochromatic signal and n(t) denotes the Gauss noise. Let the frequency of the signal f = 1 and sample the S(t) as a digital signal with IPP 25 ms and PRF 40 Hz. Select randomly a section of continual samples from the digital signal and calculate the frequency value f′ by the least mean squares linear regression algorithm. Under different SNR and N, randomly select samples with the content of the sample one hundred thousand repeatedly from the digital signal and compute the statistical distribution of f′. Figure 5 shows the distribution of probability density function of f′ under different SNR and N.Table 1 gives the statistical characteristics of f′ under various combinations of SNR and N. It can be seen from Table 1 that the precision of Doppler frequency shift will be higher and the result will be more reliable if SNR is larger and N is bigger. However, the increment of N will lead to the result that the system will spend more time at each sounding frequency. Thus it is necessary to select a suitable combination of SNR and N for the new mode of acquiring accurate Dopplionogram. On the CADI platform with the definite SNR, it is feasible for us to select N as 8 by day or 16 at night.
Table 1. Statistical Characteristics of f′ Under Various Combinations of SNR and N
Signal-to-Noise Ratio (dB)
Mean Value (Hz)
Standard Deviation (Hz)
Coefficient of Standard Deviation
Quartile Deviation (Hz)
4. The Mode for Observation of Ionospheric Disturbances and Preliminary Results
 The most common working mode for a digital ionosonde is the ionogram mode. In this mode a scan of high-frequency pulses is transmitted vertically, and when the frequencyf of the radio wave signal is equal to the plasma frequency fp in the ionosphere, the signal is reflected. By measuring the time delay of the echo reaching the receiver, the virtual height for each sounding frequency, known as an ionogram, is acquired. The new mode for observation of ionospheric disturbances in this study not only can acquire common digital ionograms, but also can acquire the Doppler frequency shift for each sounding frequency, named Dopplionogram. The measured precision of the acquired Dopplionogram in this study is better than 0.1 Hz with the time resolution 3 min, and this fully meets the demand for observation and research of ionospheric disturbances. In the following descriptions the mode for observation of ionospheric disturbances is introduced and preliminary observation results are given.
4.1. The Mode for Observation of Ionospheric Disturbances
 The new mode for observation of ionospheric disturbances has been developed using CADI digital ionosonde as the hardware platform, Windows 2000 as the operation system and Visual C++ 6.0 as the development tool. The system functions are described as follows: DDS board generates the designed pulse set and the scanned high-frequency pulse wave signal is transmitted by the transmitter; the receiver samples the echo signal data and the raw data are stored in memory buffer of PC; the stored data are analyzed, processed, and calculated in real time, the common digital ionogram and the Dopplionogram are acquired and are displayed on the monitor of PC, and the ionogram and the Doppler frequency shift data are saved in PC with the format of a permanent binary file. In addition, while the system runs, raw data can be completely saved in PC with the format of a permanent binary file so that in the future they can be used to further extract sounding information off line.
 In the mode for observation of ionospheric disturbances, the data processing algorithm is described as follows: DDS module generates high-frequency sweep frequency carriers with the frequencies from 1 MHz to 20 MHz, which are modulated by Barker 13 generated by the MCU in dual phase modulation mode; the modulated signal is transmitted by power amplifier and antenna and the echo is received by two orthogonal receiving antennas. The echo is expressed inequation (5), where ωRF is angular frequency of the radio frequency signal and Δωis Doppler frequency shift of the high-frequency signal. The local oscillating frequency signal is expressed inequation (6). The echo signal and the local oscillating signal are mixed to generate the intermediate frequency signal, which is denoted by Sout1 in equation (7). In phase and quadrature phase signal of local intermediate frequency signals are respectively listed in equations (8) and (9). Equations (10) and (11) respectively denote the demodulated results as I (I component) and Q (Q component). After the received IQ signals are sampled, quantized and correlated with Barker 13, I and Q are transported to the PC and saved in memory buffer of PC for further processing in real time. The sampling frequency of I, Q signals is 50 kHz, that is, the sampling interval is 20 μs with the height resolution of 3 km:
When system software receives the above data, a series of real-time processing is done for each of the scan frequency, for example, calculating the noise mean power, acquiring the dynamic threshold, suppressing noise by FFT transform, determining the location of the echo, separating O wave from X wave, calculating the phase value of the single narrowband echo signal, unwrapping the ambiguity of the phase and carrying out the least mean squares linear regression algorithm and so on, to acquire the Doppler frequency shift of the echo. These processing steps are described in the following:
where is the noise mean power and NT is an adjustable coefficient. The signal whose energy is larger than T is treated as effective echo.
 3. FFT analysis on the N pulses data at the same height in each sounding frequency is expressed in equations (14), (15), and (16):
where x(n) is the nth sample data at the same height for one of the sounding frequencies and X(k) is the result of FFT analysis.
 4. The three antennas are used on CADI. Twofolded dipoles for receiving, and an ‘inverted V’ for transmitting. The antenna for transmitting is mounted at north-south orientation. The receiving antennas are installed parallel to the ground in orthogonal mode, one at north-south orientation and another at east-west orientation. The echo signal data sampleSEWreceived by the receiving antenna at east-west direction andSSNreceived by the receiving antenna at south-north direction are listed inequation (17):
O Wave SO is separated from X Wave SX according to the following algorithm:
where φn is the phase of the single narrowband echo signal. In order to ensure the continuity of the phase, the arithmetic of unwrapping the ambiguity of the phase should be done.
 6. Doppler frequency shift at every height is calculated by the least mean squares linear regression algorithm in equations (21), (22), and (23):
where is the mean value of the variable φ(ti) and ti is the point of time at which the phase of the echo signal at the same height is calculated. The time interval of two adjacent tis is 25 ms on CADI. The is the mean value of the variable ti.
 According to equation (24), phase velocity of ionospheric movement can be simply obtained with Doppler frequency shift
where V* denotes phase velocity; C is the velocity of the electromagnetic wave; Δω denotes Doppler frequency shift; and f0 denotes the sounding frequency.
4.2. Preliminary Observation Results and Discussions
 In order to test the new mode for observation of ionospheric disturbances, we carried out some continual experimental observations by the newly developed mode for observation of ionospheric disturbances on the CADI platform installed at Wuhan Ionospheric Observatory (30.5°N, 114.3°E) (WIO) and obtained results shown by Figures 6–10. It should be noted that the Doppler frequency shifts in all the following figures are referenced to that of O Wave in F2 layer.
Figure 6 shows the digital ionogram and the Dopplionogram collected by the CADI digital ionosonde at 12:21 LT on 28 December 2011 at Wuhan Ionospheric Observatory. Figure 6a reveals the curve of variation of ionospheric virtual height as a function of sounding frequency, in which two traces respectively denote O Wave expressed in equation (18) in color dark gray and X Wave expressed in equation (19) in color light gray. Figure 6b shows the comparison of the Dopplionogram by FFT method with that by LMSLR method. Figure 6c shows the measured SNR of the O Wave. It can be seen from Figure 6c that SNR of the O Wave on CADI is more than 12 dB in most cases. Figure 6d give the coefficient of determination of LMSLR method. Those points of time where the value of the coefficient of determination is close to 1 are considered to possess higher reliability of Doppler frequency shift. Seen from Figure 6, it is obviously that the precision of Doppler frequency shift by LMSLR method is higher than that by FFT method and the Doppler frequency shifts by LMSLR method change with good continuity. This proves that the echo signals meet the narrowband requirement.
Figure 7shows the variations of multiple-frequency Doppler frequency shift with time collected by the new mode for observation of ionospheric disturbances on the CADI platform at Wuhan Ionospheric Observatory within one 24 h time from 27 December 2011 to 28 December 2011. The four subgraphs are from different frequencies and times, namely (Figure 7a) from 13:57 LT to 16:24 LT on 27 December 2011, (Figure 7b) from 16:54 LT to 19:24 LT on 27 December 2011, (Figure 7c) form 7:18 LT to 10:06 LT on 28 December 2011, and (Figure 7d) from 11:06 LT to 13:24 LT on 28 December 2011. The data shown in Figure 7 are replotted in Figure 8to show the contours of Doppler frequency shift of multiple frequency as a function of local time and scan frequency. The Doppler frequency shifts of the transmitted high-frequency signals are continuous as shown inFigure 8.
 It can be seen from Figures 7–9 that the Doppler frequency shift and the phase velocity continuously change with time for different sounding frequencies which correspond to different reflection heights, and the curves of the ionospheric disturbances present obvious fluctuation property and possess similar movement characteristic. The frequency spectrum analysis to the fluctuant curves in Figure 7 indicates that the quasi period of the ionospheric disturbance is approximately 30 min in Figure 7a, 16 min in Figure 7b, 60 min in Figure 7c, and 27 min in Figure 7d respectively. It is probable that the ionospheric disturbances shown in Figures 7a, 7b, and 7dare the middle-scale acoustic gravity waves whereas the disturbances shown inFigure 7care caused by large-scale acoustic gravity waves [Ding et al., 2004].
Figure 10 shows the contour of Doppler frequency shift with local time and scan frequency continually collected by the new mode on the CADI platform from 14:00 LT on 27 December 2011 to 14:00 LT on 28 December 2011 within one 24 h time. It can be also seen from Figure 10 that the top envelop of the contour is practically the critical frequency of F2 layer foF2 versus time and the bottom envelop is the lowest frequency of F2 layer versus time.
 The above experimental results also reveal that in the newly developed mode for observation of ionospheric disturbances, the system works stably and reliably, the application software for the new mode is practical and user interface friendly, and the accurate Dopplionogram and the variations of the phase velocity of the ionospheric disturbances with time and scan frequency can be continually acquired. These can commendably meet the demand for observation and research of ionospheric disturbances.
 This paper describes a newly developed mode for observation of ionospheric disturbances on the CADI ionosonde platform in vertical sounding. Some experiments have been carried out by the new mode on CADI at Wuhan Ionospheric Observatory. The main results are summarized as follows:
 1. In ionospheric vertical sounding, Doppler frequency shift can be obtained by means of measuring the variations of echo phase with time so that ionospheric disturbances information can be acquired. Instead of the traditional manner of obtaining ionospheric disturbances information by FFT frequency domain method with low accuracy and long FFT integral time, the new mode uses a combination of coded pulses and echo phase measurement analysis in time domain and therefore takes on such advanced characteristics as high accuracy and real-time operation.
 2. CADI, which is a low-cost, full-featured, flexible modern ionosonde, was introduced to China by us for the first time. Up to now, there are over six ionospheric observatories setting up CADI in China for ionospheric routine observation and scientific research. More importantly, due to its open structure and flexible operability, CADI is very suitable for system redevelopment so as to satisfy new observation mode development in a routine vertical sounding instrument.
 3. The work on CADI reveals that the new mode for observation of ionospheric disturbances is practical and feasible, and the system works stably and reliably. Moreover, the new mode of acquiring the Dopplionogram with high accuracy in real time can commendably work without interrupting CADI's routine ionogram mode. The realization of the new mode can be taken as a useful supplement to the traditional vertical sounding methods and can also be easily applied to other ionospheric vertical sounding systems. This not only opens up a new effective way by which much more ionospheric observation information can be acquired with the existing routine ionospheric sounding instrument, but also provides many rich experiences in developing more new ionospheric vertical sounding modes and extending the functions of digital ionosondes.
 However, it must be pointed out that this method needs the condition that the echo is a narrowband and single-frequency signal with proper SNR. Although in most cases this condition can be satisfied, the measurement precision may be lower at night because of lower signal-to-noise ratio then. This problem can be solved to some extent by increasing the numberN of the transmitted pulses at the same sounding frequency. For our next work, we would invert the profile of ionospheric disturbances with the obtained accurate Dopplionograms and extract ionospheric disturbance characteristics. At the same time, we would measure ionospheric virtual height by analyzing the variations of the echo phase with the transmitted frequency, and the measurement precision of the virtual height can most likely be improved. This would greatly contribute to ionospheric movement and subtle structure research.
 This work is supported by Natural Science Foundation of Hubei Province, China (2010CDA062), National Science and Technology Basic Work Program (2008FY120100), and National Natural Science Foundation of China (41127003). The authors would like to thank the reviewers for their constructive comments and suggestions to improve the paper.