Very High Frequency (VHF) radar provides an important method for stealth target detection with the target resonance effect which increases Radar Cross Section (RCS) significantly. However, the traditional narrowband VHF radar has the disadvantages of being difficult to determine optimum operating frequencies, unstable performance in low attitude, being vulnerable to interference, and so on. Wideband technology provides a way to overcome the above disadvantages. The target echo of wideband VHF radar is different from that of wideband radar in the optical region, Boeing747-200 is selected as the specific object to research wideband VHF radar target echo model and its detection performance. First, the simulated wideband echo data are obtained, and then the multiple scattering centers’ model is established based on the simulated data. Finally, the detection performance of wideband VHF radar is discussed. Results show that in the resonance region targets have the property of multiple scattering centers; the main peak intensity of the scattering center is equal to the mean value of the echo intensity of the narrowband signals in the working band; wideband VHF radar does not have the obvious detection advantage compared with the traditional narrowband VHF radar. This investigation establishes the theoretical foundation for the design of novel wideband VHF radar system with the capability of detecting stealth targets.

Sidelobe clutter suppression is an important issue in target detection. In this paper, a beampattern design method based on a priori information is proposed to suppress clutter in sidelobe. Based on the correlation matrix of the echo in the interesting region with transmitted signal being orthogonal waveform, the cost function for correlation matrix of transmitted waveform is established by employing the maximum Signal-to-Clutter-plus- Noise Ratio (SCNR) criterion, and it can be solved by Semi-Definate Programming (SDP). Then, the transmitted waveform is optimized by using conjugate gradient. The numerical results show that the optimized waveform can increase the SCNR in the echos efficiently, in the non-homogeneous environment with strong clutter.

Because the real orthogonal waveforms occupying the same frequency band may not realistically be found, the imaging quality is degraded by the cross-correlation noise caused by coupling among these waveforms for MIMO-SAR system that transmits the same frequency coded orthogonal waveforms. To address the above problem, a novel orthogonal waveform separation method is proposed based on Clean processing and is used to obtain the echoes from each waveform. Then, its performance is discussed. Finally, the numerical experiments are provided to demonstrate the effectiveness of the proposed method to suppress the cross-correlation noise among orthogonal waveforms for point scatterer and verify that its imaging performance approaches that of conventional SAR with single transmitting waveform. 

Low sidelobe transmit pattern of MIMO radar can concentrate majority energy in the mainlobe region to improve the target detection, and reduce the energy of clutter and false targets from sidelobe, so as to enhance the SNR of echo. Based on a modified non-diagonal elements of transmit signal covariance matrix, a low sidelobe transmit pattern optimization algorithm for MIMO radar is proposed. Firstly, the pattern matching design method is implemented to obtain the desired mainlobe shape of transmit pattern, with the results as the initial value. Secondly, the cost function can be formulated as the minimization of peak-sidelobe or integrated-sideloble; the mainlobe shape distortion can be constrained by the Frobenius norm of modification matrix, and the modified transmit signal covariance matrix needs to be positive semi-definite. Since the two optimization problems are convex, the global optima can be acquired. The simulation results show that the desired mainlobe shape and low sidelobe pattern can be obtaibed with the proposed method.

Nadir echo and the echo from the range ambiguity points will produce serious interference to SAR image. To deal with this problem, a method which can avoid the influence of these useless signal is proposed in this paper. This method is based on the Digital Beam Forming (DBF) theory while the received echoes are weighted in real-time and the ambiguity points echo will be eliminated while maintaining the useful signal at the same time. This method can also provide more possibilities for the selection of Pulse Repetition Frequency (PRF) and reduce the difficulty of the spaceborne SAR system design. The simulation results with the point targets validate the effectiveness of the proposed method.

Skywave Over-The-Horizon Radar (OTHR) can achieve long range detection via the ionosphere reflection. However, time-varying ionosphere and multipath effect will lead to serious clutter spectrum spread in long Coherent Integration Time (CIT). It increases difficulty degree of the detection of slow-speed targets. To deal with contaminated echo in the case of the two spreading Bragg peaks overlapping, an ionospheric contamination correction scheme based on Short Time MUSIC Transform (STMT) and improved Phase Gradient Autofocus (PGA) method is proposed. In the scheme, coarse estimation of ionospheric phase contamination can be obtained and compensated by using STMT transform, then an adaptive sliding window filter is utilized to extract the first order clutter (calibration signal). After that, the remanent ionospheric phase contamination can be estimated and compensated accurately by using improved PGA method. Theoretical analysis and simulation results show that the scheme has a good performance on correcting the situation of two spreading Bragg peaks overlapping.

The low-frequency SAR is always easily interfered by other wireless services such as mobile communications, television, which operate in the same band. According to the narrow band of the Radio Frequency Interference (RFI) signals, this paper proposes a RFI suppression method based on subband space projection filtering. This method sorts and recognizes the interfered subband first and then suppresses the RFI with the space projection filtering. The proposed method can suppress the RFI effectively. Both the simulation and experimental results are provided to illustrate the effectiveness of the proposed approach.

When the sliding spotlight mode SAR works in spaceborne squint case or airborne high-squint high- resolution case, the coupling between range and azimuth rises rapidly. It leads to a new two-dimensional (2-D) spectrum skew phenomenon. The skew extend the azimuth spectrum and enlarge the Doppler bandwidth, which causes a new azimuth spectral folding effect. The folding can not be eliminated by the conventional two-step processing algorithm. In this paper, a modified two-step processing algorithm is proposed. Efficiently implemented without large zero padding, it eliminates the effect of the 2-D spectrum skew. Therefore, the algorithm can process the data of squint sliding spotlight mode efficiently. The processing flow of the algorithm is described in detail, and compared with the conventional two-step processing algorithm. Finally, experiments on simulated and real data validate the effectiveness of the proposed algorithm in squint sliding spotlight case.

In this paper, the effects on signal characteristic which caused by cubic range migration in high squint mode are considered, and a modified chirp scaling imaging algorithm is proposed. In this method, range walk is removed in time domain firstly, and range cell migrations of all targets in the imaging scene are corrected accurately by chirp scaling operation and the unified Range Cell Migration Correction (RCMC). Finally, a focused SAR image can be obtained after the modified azimuth filtering. The overall imaging procedure of the algorithm only involves fast Fourier transform and complex multiplication, which means higher efficiency. Simulation results show that the modified algorithm can greatly improve the azimuth side-lobe performance, which is a great advantage in the high resolution imaging in high squint mode.

Forward looking missile borne SAR suffers the issue of left/right ambiguity in SAR images because the targets situated symmetrically about the moving trace of radar have the same Doppler histories. Considering this issue, the imaging model of double antenna forward looking SAR is established based on the characteristics of the missile borne SAR, and the signal features in the double antenna system are analyzed. Based on the above analysis, a scheme is put forward to solve the left/right ambiguity. Firstly, two ambiguous SAR images are obtained with squint SAR imaging algorithm. Then the left/right ambiguity is solved by beam-forming based on range delay difference between different antennas. Finally, the point target imaging simulation testifies the validity of the proposed algorithm.

Ground and sea echoes with anomalous propagation are serious issues affecting Doppler radar data quality. Based on the weather radar equation and the propagation model for tropospheric duct, a method of calculating the ground and sea echoes of weather radar is presented. With refractivity profiles and digital elevation data as input, the method can calculate the distribution of the path loss. Then the reflectivity fields can be obtained considering the parameters of the radar and the scatting coefficient of the earth surface. The ground and sea echoes are simulated in standard and atmospheric duct environments for Doppler weather radar located at the coast of Bohai, which indicate the effectiveness of the method in simulating the anomalous echoes.

A recognition and parameter estimation algorithm for hybrid modulation signal combined with Frequency Shift Keying and Binary Phase Shift Keying (FSK/BPSK) is proposed. First, the models of the considered signals and their unwrapped phase are presented. Then, a recognition algorithm is presented for signals including FSK/BPSK hybrid modulation signal, BPSK signal and FSK signal based on instantaneous frequency by utilizing binary tree, and the thresholds of recognition based on Neyman-Pearson (N-P) criterion are discussed and analyzed in detail. Finally, the parameter estimation of FSK/BPSK signal is implemented. Simulation results indicate that the proposed algorithm possesses fine recognition performance and estimation accuracy in lower SNR levels.

Repeater jamming is an important trend of deception jamming against Linear Frequency Modulation (LFM) radar, and repeater jamming waveform is determined by three factors: carrier frequency offset, pulse width and linear frequency modulation rate. When there are simultaneous deviations of these factors, the deception effect of repeater jamming is studied separately in time domain and frequency domain. Three important parameters about Jamming to Noise Ratio (JNR) are derived especially, which are pulse width losses, peak power losses and peak time delay. Theoretical analysis shows that the fast estimation by a group delay in time domain for the formulas of these parameters is in accordance with the detailed deduction by spectrum analysis in frequency domain,  and the corresponding MATLAB simulations are provided. So the analysis of the general jamming can be simplified by these formulas.

Due to the issues in multichannel interferometer direction finding, such as system complexity and demanding geometry configuration requirement for ambiguity resolving, a Pseudo-linear Least Square (PLS) algorithm is proposed for virtual Two-Dimension (2D) interferometer using the time varying long baseline. It first obtains a set of initial values using a pair of measured phase differences, then applies PLS to each initial angle and finally picks out the estimate with the minimum cost function as the ultimate result. The equivalent relationship between Cramer-Rao Lower Bound (CRLB) and Mean Square Error (MSE) is also derived. The proposed method has no special requirement for the geometry configuration and no need of phase difference unwrapping. Simulation results show this method achieves moderate computation cost and the estimate precision can approach to the CRLB.

A method for GPS multipath mitigation is proposed. The general process of the proposed method is that first initial estimations of time delay and amplitude of each multipath are obtained by compressed sampling of the received signal with Teager-Kaiser (TK) operator, and then further estimations of the initial estimated results are got by reconstructing the received signal’s auto-correlation curve. Finally, more precise estimates of amplitude and time delay of the Line-Of-Sight (LOS) signal can be acquired from the previous reconstruction results, and thus the goal of GPS multipath mitigation is achieved. The simulation results under different conditions show that the proposed method outperforms the narrow correlator which is often used in a conventional GPS receiver, and it can also achieve a superior performance as the high resolution correlator can when one sample interval modification is done on the TK time delay estimation results.

With the development of information technology, people tend to deal with large amount of data with complex data types. Therefore, how to deal with these data well to achieve good classification results is a highly challenging task. In this paper, a new hybrid algorithm ROF-ELM is proposed, which both combines the advantages of ROtation Forest (ROF) algorithm and Extreme Learning Machine (ELM) neural network. The new algorithm solves the over-fitting problem of the original Rotation Forest algorithm and improves the classification accuracy as well. The proposed ROF-ELM algorithm is experimented on the UCI data sets and remote sensing image. Compared to classical ensemble algorithm, ROF-ELM improves the classification accuracy, and has high stability and generalization performance at the same time.

For the curse of dimensionality encountered in solving the planning in Partially Observable Markov Decision Processes (POMDP), this paper presents a novel approach to compress belief states space using Non- negative Matrix Factorization (NMF) updating rules, which reduces high dimensional belief states space by two steps. First, the algorithm adopts factored representations of states, observations and actions by exploiting the structure of factored POMDP, then decomposes and compresses transition functions by exploiting conditional independence of dynamic Bayesian network, and then removes the zero probability to lower the sparsity of belief states space. Second, it adopts value-directed compression approach to make the obtained approximate belief states after dimension reduction be consistent with the original optimal, and exploits NMF updating rules instead of Krylov iterations to accelerate the dimension reduction. The proposed algorithm not only guarantees the value function and reward function of the belief states unchanged after reducing dimensions, but also keeps the piecewise linear and convex property to compute the optimal policy by using dynamic programming. Experiments demonstrate that the proposed belief compression algorithm has lower error rates and higher convergence.

A distinct non-local means denoising algorithm derived from structure-adapted block matching is proposed in this paper. Multi-scale matching of image blocks is adopted to measure similarity of local structures, which can deal with complex structural characteristics effectively and subsequently improve denoising performance. To begin with, structural region (including edges and textures) and flat region are divided by introducing Coefficient of Variation (CV) characteristics and the CV-Kmeans region classification algorithm is proposed. Furthermore, the size of similar block is adaptively selected based on average Euclidean distance between blocks in structural regions. Finally, a new non-local means algorithm is proposed to remove noise. Compared to the classical non-local means algorithm, the improved algorithm using patch probabilistic similarity and the adapted non-local means denoising algorithm, experimental results show that the proposed algorithm increases denoising performance and especially demonstrates a distinct advantage in texture images.

In order to preserve the geometric characteristics of image, a weighted wavelet variational inpainting model is proposed. The new model makes use of the advantages of weighted function that can improve the capacity for denoising in the smooth region and protect the edges in the edge region, so it can enhance the inpainting effect. To solve the proposed model effectively, it is firstly turned into two submodels using alternating direction method, then the corresponding theoretical derivation and algorithms are given. The numerical experiments show that the new model can not only get the better visual effect, but also obtain the higher Signal to Noise Ratio (SNR) than the recent total variation wavelet inpainting method. 

A fining algorithm of correlation peak is proposed for measuring group delay of Radio Frequency (RF) systems to improve the measurement precision. This algorithm requires lower sampling rate by interpolating the signals before correlation operations, and obtains higher measurement precision than classic methods. Wavelet de-noising is also introduced to enhance the robustness under low Signal-to-Noise Ratio (SNR). Simulations based on matlab show that the proposed method can obtain the delay part less than the sampling interval more precisely, and prove its effectiveness and advantages.

To overcome the influence of illumination and improve the accuracy of optical flow estimation for color images, a new optical flow estimation algorithm is presented, which fuses optical flow calculated in the color invariant space and that in the RGB space. First, the color invariant edges of the two adjacent frames in the video sequence are extracted. Then, the three-channel color images are built with the color invariant edges, and the optical flow of the color images is computed. Finally, the computed optical flow of color invariants is fused with that of the RGB space based on the L_∞  norm. How the fused optical flow estimation can be influenced by different selection and combination of color invariant edges are also discussed. Experimental results show that the proposed method captures object areas more accurately, and is more robust to the shadow and illumination variation than the typical algorithms.

Conventional techniques of passive video forensics are only effective for some specific tampering operations, and just provide the binary judgment about its trustworthiness. Compared with tampering detection, the processing history estimation is a deeper goal of forensics investigation. In this paper, a digital video tampering forensics scheme is proposed based on forensics hash. To estimate the processing history, it analyzes the artifacts of video tampering, and the statistical features of several video tampering such as adding/deleting motion object are extracted as side information. Forensics hash components are constructed for geometric transform estimation, adding/deleting video object and tampering localization. A video forensic hash platform is built in a modular way. Experimental results show that the proposed approach achieves the processing history estimation of many video tampering operations such as geometric transform, seam-carving based video retargeting, and object-based manipulation. Moreover, it realizes rough tampering localization with a relatively low computational complexity.

Due to the time-varying anatomical information in the applications of magnetic resonance dynamic imaging and functional neuroimaging, the traditional constrained imaging suffers the imaging error and the long registration time. In this paper an imaging strategy is proposed to reconstruct rapidly a low-resolution image as the reference image, and make the image registration based on Mutual Information (MI) metric of two images combined with landmark curvature mapping. The proposed method calculates and compares the MI of adjacent frames of temporal image sequences before and after registration to decide whether to update reference images, as well as to verify the registration results. Experimental results show that the proposed method achieves high registration precision and reduces dramatically the computational cost, and thus it can be applied to real-time dynamic imaging.

In order to correct the signal time-difference of arrival estimation errors introduced by correlation peaks ambiguity in underwater acoustic passive localization, an ambiguity elimination algorithm is proposed based on the propagation characteristics of underwater sound. The physical essence of correlation peaks ambiguity is studied using the ray model of underwater acoustic propagation. Distribution disciplines are then summed up for correlation peaks caused by multi-path propagation in shallow sea acoustic channel. The amplitude and time-delay characteristics of correlation peaks between different paths are analyzed by simulations, and the ambiguity elimination algorithm is proposed based on correlation peaks classification and stable peaks tracing. The algorithm is used to estimate the time-difference of arrival for signal acquired in sea trial. Processing results show that estimation errors can be effectively corrected with the proposed algorithm.

An improved Propagator Method (PM) is proposed for joint Time-Of-Arrival (TOA) and Direction- Of-Arrival (DOA) estimation in Impulse Radio Ultra WideBand (IR-UWB) system. After modeling the received signal in frequency domain, the proposed algorithm estimates the TOA parameters in two antennas, and obtains the DOA parameters via the difference of the TOAs between the two antennas, thereby gains the joint TOA and DOA estimation. The proposed algorithm can obtain the closed-form solutions of the TOA and DOA parameters without spectral peak searching and these solutions can automatically get paired, thus significantly reduces the complexity. An amount of simulation experiments about the proposed algorithm and the comparisons with the existing algorithms are given. Simulation results verify the practicability, robustness and superiority of the proposed algorithm.

In order to mitigate the Narrow-Band Interference (NBI) signal in Direct Sequence Spread Spectrum (DSSS) system, a Difference-Cluster-Threshold (D-CT) narrowband interference suppression algorithm based on threshold detection in transform domain is proposed. The D-CT algorithm takes advantage of the difference values of the spectrum envelope of the received signal to find the non-interference signal spectrum segments, and thus it can suppress the NBI signals by using the maximum amplitude modulus value of the spectrum segment as the threshold. Statistical analysis shows that the number of multiplications of the adaptive multi-threshold algorithm is about 5.4 times that of the D-CT algorithm. The number of additions of the D-CT is only about one fourth that of the adaptive multi-threshold and one fifth that of the Double Threshold Forward Consecutive Mean Excision (DT-FCME) respectively. Simulation results indicate that the simulation time of the D-CT algorithm is approximately one third that of the adaptive multi-threshold algorithm and one fourth that of the DT-FCME algorithm respectively without losing the BER performance.

The projection of multi-antenna weighted vector structured with spatial scrambling is of constant modulus on the legitimate channel. Eavesdropper can take the advantage of multi-antenna to intercept private information. To solve this issue, a kind of physical-layer security algorithm is designed based on the joint scrambling with spatial and frequency resource. Firstly, random multi-antenna weighted vector is constructed in each subcarrier, and the vectors projection on the legitimate channel is equal to a reference variable that changes randomly in each OFDM symbol. Secondly, the weighted vector delivering the reference variable is transmitted in the reference subcarrier, with which the legitimate user can correctly demodulate the remaining subcarriers. Whereas, eavesdropper can not obtain the useful information because of the nature difference between the legitimate channel and eavesdropper channel. The simulation results show that, when SNR achieves 10 dB, the legitimate user’s Bit Eerror Rate (BER) is 10^-5 , the eavesdropper can not intercept the private information using blind equalization or MUSIC-like method.

With the development of mobile network, there is a stricter requirement for the quality of personalized mobile user services. In order to obtain more accurate mobile user preferences, a prediction method of mobile user preferences based on trust and link prediction is proposed by analyzing mobile user behaviors. Firstly, a trust calculation method of mobile user is proposed according to the mobile users’ communication behaviors. Then the nearest neighbors of active user are chosen with the obtained trust and the similarity of the ranking to the mobile network services. Then the relate degree between the mobile user and the mobile network service is obtained by the link prediction, and the mobile network services that need to be predicted can be ascertained by the obtained relate degree. Finally, the experimental results show that the proposed method can obtain more accurate mobile user preferences compared with the traditional collaborative filtering, and it can solve the sparsity issue to some extent.

Fault in a rearrangeable multistage multi-plane Clos optical switching fabric network will change the non-blocking characteristics of the switching fabric and introduce traffic blockage. Based on analyzing the properties of the faults, a threshold-based limiting algorithm is proposed. Considering the combined fault effects to different layers, the algorithm finds out the traffic loads which can not be supported due to the fault. Those traffic loads are then limited in the outmost layer to achieve efficient scheduling of the following layers and to reduce the blocking probability. Simulation results show that the limiting algorithm can reduce the blocking probability effectively. With the deepening of the fault location and time slot increasing, the algorithm has a better performance.

Traditional packet classification algorithms often have many redundant rules. To solve this issue, a packet classification algorithm called Clustered Dynamic Point Split (CDPS) is proposed based on the analysis of the characteristics of rule sets. CDPS divides the rule set by clustering the rules with similar cross-space relationship, then, it dynamically selects the rule projection points to complete the space decomposition and to build the decision tree. Simulation results show that, without reducing the time performance, the memory cost of CDPS is 95% and 50% less than HyperSplit and EffiCuts, respectively.

As indoor wireless sensor networks channel fading resulting in frequent topological changes, an adaptive routing algorithm is necessary with the changes of environment. To this end, an Energy-Balanced Adaptive Clustering Routing (EBACR) algorithm is proposed in this paper.  In this algorithm, a cluster head node and its relay node are chosen by a comprehensive consideration of the Receiver Signal Strength Indicator (RSSI) values received by sensor nodes and the residual energy of sensor nodes, and the other nodes can then decide which cluster they will belong to. EBACR algorithm can adaptively build the routing by multiple attribute decisions, according to the changes of different attributes. The experimental results show that EBACR algorithm can get better performance in the aspect of energy balance and network lifetime compared with other three algorithms.

Fully Homomorphic Encryption (FHE) makes it possible to perform all sorts of calculation in cipher domain, and own great value in cloud computing. However, the security of existing FHE schemes under the non-adaptive chosen-ciphertext attacks remains an issue to be studied. In this paper the construction of the Learning With Errors-based FHE scheme is analyzed, and it is pointed out that this cryptosystem has a potential security flaw, which creates the risk of private key leakage when confronting non-adaptive chosen cipher attacks. According to this, a key recovery attack is proposed, which is capable of recovering the private keys of such cryptosystem with the help of decryption oracle, through blending the dichotomy approximation in cipher domain and the solution of linear congruent equations.

Compiling speed and code quality are two key factors to evaluate the performance of register allocation. Modern Just In Time (JIT) compilers desire the best code in least time. Neither traditional graph coloring algorithm nor linear scan algorithm can satisfy the condition perfectly. A regions priority algorithm is proposed in this paper, which does not pursue the perfect register allocation in theory but takes greedy design method. The program is divided by loop region, on which physical register allocation is implemented by dealing lifetime length priority queue and spill weight priority queue. Then the allocation extends to the whole program. This algorithm creates high quality code in linear time and is applied to the compiler which compiles PTX instructions in SSA to traditional multicore platform. The experiment based on this compiler verifies the effectiveness of the algorithm.

An analytical model for the parasitic capacitance of tapered Through-Silicon-Vias (TSV) with MOS effect is proposed by solving Poisson’s equation. A comparison between the analytical model and Ansoft Q3D parameter extraction model is given based on copper TSV. The results show that in the bias voltage of -0.4 V, 0.5 V and 1.0 V, for the tapered TSV slop wall angles of 75°, 80°, 85° and 90°, the maximum Root Mean Square (RMS) errors of analytical model are respectively 6.12%, 4.37%, 3.34% and 4.84% over a wide range of multiple parameters; when MOS effect is ignored, the maximum RMS errors are respectively 210.42%, 214.81%, 214.52% and 211.47%, and it proves that the analytical model is accurate and the consideration of MOS effect is necessary to the analytical model. Taking into account the MOS effect, the maximum damping of S₁₁ and the maximum increase of S₂₁ are about 19 dB and 0.01 dB respectively, simulated by Ansoft HFSS, so the transmission performance of tapered TSV is improved.

A Dynamic Width Controller (DWC) for switch mode DC-DC converter is designed. Based on the comparison result of the sensed load current and the internal reference current, the controller can select the optimum switch width and drivers for the DC-DC converter according to the load current condition. The controller can improve the efficiency of DC-DC converter for high switching frequency and light load applications. The chip adopts a low power cost Voltage-Current Converter (VCC) to provide load current information for the DWC. To ensure the security of the chip’s performance, protection circuits are also integrated. The chip is simulated and fabricated under standard 0.35 μm CMOS process. The experiment results show that the proposed circuit can achieve dynamic switch width control for the DC-DC converter, and the maximum efficiency improvement is 6% for light load condition.

A two-level arbitration algorithm based on weight and Round-Robin (RR) is presented. It sets weight by tickets and employs improved Fixed Priority (FP) and RR arbitration to work in turn respectively under the conditions that there is no contention and there exit heavy contentions. In the NonIdling and NonPreemptive (NINP) model, the proposed arbitration algorithm is much better in the output bandwidth ratio, bandwidth utilization, power, fanout and it also has advantages in speed and area compared with the commonly-used FP, RR and Lottery arbitration algorithms. The proposed arbitration algorithm is suitable in various request environments, simple in logic and easy to implement, so it can be applied to SoC bus systems.

A novel and high-performance micromachined electric-field sensor system circuit which can measure both AC (50 Hz) and DC fields is presented based on Direct Digital Synthesizer (DDS) and Lock-In-Amplifier (LIA). The system combines low-noise analog circuits and digital control scheme, using DDS to generate the excitation signal and the orthogonal reference signals, ensures the frequency and amplitude stability of the excitation signal and precise phase difference between the orthogonal reference signals. Using demodulator chip and high-speed filter to build quadrature lock-in amplifier circuit, combining with the ARM microcontroller, the system achieves a faster response and saves much more sources than the simple digital processing circuit.

The OFDM waveform employed in SAR systems is a novel waveform for Ultra-Wideband (UWB) SAR, which attracts many attentions in recently years. In this paper, the influence of Doppler shift on OFDM SAR is analyzed in terms of ambiguity function. The Doppler shift can not only reduce the pulse compression coefficient of OFDM signals in range, but also affect the focusing quality of SAR systems in azimuth. In order to overcome the restriction of Doppler shift on OFDM SAR systems, a method of Doppler compensation in range Doppler domain is designed. Simulation results demonstrate the effectiveness of the compensation method.

In order to improve the positioning accuracy of near-field target, a passive location method is proposed by measuring the phase difference, which can be mapped into the time difference after solving the near-field fuzzy phase problem, and then a near filed location model is proposed based on phase measurement. The location model consists of nearby and remote sensor array that is connected with the central processing unit through optical fiber, thus the whole configuration works as a distributed measurement system. By overcoming the loss and fuzzy phase problem, then time difference can be obtained through phase difference indirectly. Such configuration could achieve accurate and quick location of the radiation source. Simulation results indicate that the proposed approach offers higher accuracy, lower loss and simplified large-area construction compared with the location scheme based on time-difference measurements. Therefore, it may find values in the field of electronic countermeasures.

The recommender algorithm based on Regularized Singular Value Decomposition (RSVD) has significant advantages in predictive accuracy, while it is computationally intensive, which limits greatly its application to engineering projects. To address this issue, an improved algorithm based on tag transfer learning is proposed. It leverages tag information in the relatively denser auxiliary dataset to extract user/item features, which are further used in the RSVD approach in order to make recommendation in the target dataset. Experiments on MovieLens datasets show that the proposed algorithm can handle the sparsity issue effectively, achieve far better prediction results (reducing about 0.01 RMSE), and save about 50% training time at the same time.