Accelerate Your 3D X-ray Failure Analysis by Deep Learning High Resolution Reconstruction

Over the past decade, 3D X-ray technique has played a critical role in semiconductor package failure analysis (FA), primarily owing to its non-destructive nature and high resolution capability [1,2]. As novel complex IC packages soar in recent years [3,4], X-ray failure analysis faces increasing challenges in imaging new advanced packages because IC interconnects are more densely packed in larger platforms. It takes several hours to overnight to image fault regions at high resolution or the crucial details of a defect remain undetected. A high-productivity X-ray solution is required to substantially speed up data acquisition while maintaining image quality. In this paper, we propose a new deep learning high-resolution reconstruction (DLHRR) method, capable of speeding up data acquisition by at least a factor of four through the implementation of pretrained neural networks. We will demonstrate that DLHRR extracts signals from low-dose data more efficiently than the conventional Feldkamp-Davis-Kress (FDK) method, which is sensitive to noise and prone to the aliasing image artifacts. Several semiconductor packages and a commercial smartwatch battery module will be analyzed using the proposed technique. Up to 10x scan throughput improvement was demonstrated on a commercial IC package. Without the need of any additional X-ray beam-line hardware, the proposed method can provide a viable and affordable solution to turbocharge X-ray failure analysis.

Throughput Improvement in Femtosecond Laser Ablation of Nickel by Double Pulses

In this study, femtosecond laser double pulses were tested to improve their nickel ablation efficiency. The experimental results indicated that compared with single pulses, double pulses with different delay times generated craters with larger diameters and depths. The results obtained for three sets of double pulses with different energy ratios indicated that double pulses with an energy ratio of 1:9 had the highest ablation efficiency, followed by those with energy ratios of 2:8 and 5:5. The double pulses with the aforementioned three energy ratios achieved the maximum ablation efficiency when the delay time was 3–4 ps. Compared with single pulses, double pulses with an energy ratio of 1:9 generated craters with an up to 34% greater depth and up to 14% larger diameter. In addition, an interference effect was observed with a double pulse delay time of 0 ps, which has seldom been reported in the literature. The double pulses were simulated using the two-temperature model. The simulation results indicated that double pulses with an energy ratio of 1:9 with a delay time of 4 ps can perform the strongest ablation. These simulation results are in line with the experimental results.

A Power and Spectrum Efficient Uplink Transmission Scheme for QoS-Constrained IoT Networks

<div>Abstract—Non-orthogonal multiplexing (NOM) is a novel superposition coding inspired scheme that has been recently proposed for improving the power, spectrum efficiency and delay of wireless links with packet error rate (PER) constraints. Despite its efficiency, restricting the number of multiplexed packets to two limits the throughput improvement to 100%. Therefore, this work presents a novel NOM design with unlimited number of multiplexed packets by manipulating the repeated transmissions in automatic repeat request (ARQ) to enhance the power and spectrum efficiency by multiplexing new and repeated packets while taking into account the channel conditions and varying the power per packet in different transmissions. The proposed scheme employs an efficient heuristic algorithm to perform the power assignment and multiplexing decisions. Moreover, the complexity of the proposed NOM can be controlled by enforcing a limit on the maximum number of multiplexed packets per transmission, making it suitable for different types of Internet of Things (IoT) nodes with various computational capabilities. The obtained results demonstrate the effectiveness of proposed scheme, which offers up to 200% spectral efficiency improvement at moderate signal to noise ratios (SNRs), and up to 700% at high SNRs. Furthermore, the new scheme can reduce the transmission power consumption by up to 6 dB in the high SNR region.</div>

A Power and Spectrum Efficient Uplink Transmission Scheme for QoS-Constrained IoT Networks

<div>Abstract—Non-orthogonal multiplexing (NOM) is a novel superposition coding inspired scheme that has been recently proposed for improving the power, spectrum efficiency and delay of wireless links with packet error rate (PER) constraints. Despite its efficiency, restricting the number of multiplexed packets to two limits the throughput improvement to 100%. Therefore, this work presents a novel NOM design with unlimited number of multiplexed packets by manipulating the repeated transmissions in automatic repeat request (ARQ) to enhance the power and spectrum efficiency by multiplexing new and repeated packets while taking into account the channel conditions and varying the power per packet in different transmissions. The proposed scheme employs an efficient heuristic algorithm to perform the power assignment and multiplexing decisions. Moreover, the complexity of the proposed NOM can be controlled by enforcing a limit on the maximum number of multiplexed packets per transmission, making it suitable for different types of Internet of Things (IoT) nodes with various computational capabilities. The obtained results demonstrate the effectiveness of proposed scheme, which offers up to 200% spectral efficiency improvement at moderate signal to noise ratios (SNRs), and up to 700% at high SNRs. Furthermore, the new scheme can reduce the transmission power consumption by up to 6 dB in the high SNR region.</div>

Data Transmission Evaluation and Allocation Mechanism of the Optimal Routing Path: An Asynchronous Advantage Actor-Critic (A3C) Approach

The delay tolerant networks (DTN), which have special features, differ from the traditional networks and always encounter frequent disruptions in the process of transmission. In order to transmit data in DTN, lots of routing algorithms have been proposed, like “Minimum Expected Delay,” “Earliest Delivery,” and “Epidemic,” but all the above algorithms have not taken into account the buffer management and memory usage. With the development of intelligent algorithms, Deep Reinforcement Learning (DRL) algorithm can better adapt to the above network transmission. In this paper, we firstly build optimal models based on different scenarios so as to jointly consider the behaviors and the buffer of the communication nodes, aiming to ameliorate the process of the data transmission; then, we applied the Deep Q-learning Network (DQN) and Advantage Actor-Critic (A3C) approaches in different scenarios, intending to obtain end-to-end optimal paths of services and improve the transmission performance. In the end, we compared algorithms over different parameters and find that the models build in different scenarios can achieve 30% end-to-end delay decline and 80% throughput improvement, which show that our algorithms applied in are effective and the results are reliable.

Video Compression Based on FPGA Using SIMD Architecture

Recently, video files and images have became the dominant media material for transmitting or storing across different applications that are used by different people. So, there was a serious need to find more effective and efficient video compression techniques to reduce the large size of such multimedia files. This paper proposes SIMD based FPGA lossless JPEG video compression system with the facility of scalability. Generally, the proposed system consists of a software side and a hardware side. The digital video file is prepared to be processed by the hardware side frame by frame on the software side. The hardware side is proposed to consist of two main processing circuits, which are the prediction circuit for calculating the predicted value of each pixel in the certain frame and the encoding circuit that was represented by a modified RLE (Run-Length-Encoder) to encode the result obtained through subtracting the predicted value from the real value for each pixel to produce the final compressed video file. The compression ratio obtained for the proposed system is equal to 1.7493. The throughput improvement for the two and four processing units basing on SIMD architecture was 100 MP/s and 200 MP/s, respectively. The clock results showed that the number of clocks required had become 50% and 25% when using two processing units and four processing units, respectively, from the number of clocks using single processing units. Index Terms— Video Compression, Lossless JPEG, RLE, FPGA.