An intelligent antenna synthesis method is proposed to automatically select suitable antenna type and provide optimal geometric parameters according to the requirement of antenna performance. The performance characteristics of different antennas are analyzed, and a smart antenna selection model is established to recommend the appropriate antenna by using support vector machine (SVM). Then, the stacking ensemble model is built by combining five primary learners to give full play to the advantages of each type of learners and, finally, the geometric parameters of the antenna are provided. Microstrip patch antenna, spiral antenna, and horn antenna are used to demonstrate the accuracy and efficiency of the model. The system shows a great ability of antenna classification with accuracy over 99% and parameter prediction with a mean absolute percentage error (MAPE) of less than 6%. Moreover, the results are compared with the traditional data-driven model, and the key parameters of the model are comprehensively discussed. The proposed method could be widely used in intelligent antenna design in practice.

The explosive growth of electronic devices brings a soaring demand for rapid electromagnetic compatibility (EMC) diagnosis. However, there is a significant learning curve for the electrical engineers to apply EMC knowledge. In this article, an efficient EMC diagnosis and management methodology was proposed, which provided a fast way for EMC analysis in seconds other than traditional simulation or calculation. The approach organized the EMC knowledges as knowledge graph composed by the interference/sensitive units, and mathematical set rules. The optimized graph structure is in form of rule, maxterm, basic unit, and entity layers. Based on the condensed relationships, it achieved high searching efficiency and graph expansibility. To facilitate the information retrieval from the knowledge graph, the interference/sensitive units and related parameters were acquired from interactive sessions, in which long short-term memory method was used to extract entities. The EMC specialized corpora were fed in training to enhance the accuracy of inference. Finally, the EMC diagnosis and management reports were automatically generated by knowledge graph searching application. The proposed method improved the calculation efficiency by three times. The storage of relationships and attributes of nodes was reduced by 76% and 60.7%. The identification accuracy was enhanced from 77.7% to 99.5%. The presented method is practically useful for EMC design in crosstalk analysis, radiated, and conducted interference diagnoses.

A 6-bit time-to-digital converter that achieves mismatch free operation by using a single delay cell and sampling flip-flop is presented. The proposed TDC was integrated in a digital fractional-N PLL fabricated in a 32-nm digital SoC CMOS process for WiFi/WiMax radios. The TDC consumes 3 mW from a 1.05-V supply and occupies an area of 0.004 mm². A digital frequency-locked loop is used to track and correct for PVT variations in the TDC and no additional linearization or mismatch calibrations are required. The DPLL uses a 20-bit high dynamic range DAC to drive a VCO in order to effectively realize a DCO with 100-Hz frequency resolution. The 2.5-GHz WiFi band LO output is generated from a 40-MHz reference with an integrated phase noise of - 35 dBc (10 kHz to 10 MHz) while consuming 21 mW . The worst case spur in the LO output is below - 50 dBc without requiring TDC mismatch and linearity calibration.

This article introduces a novel crosstalk prediction method, defines a standard data format for printed circuit board (PCB) modeling, and establishes an efficient computation model utilizing machine learning. The crosstalk prediction is undertaken by using a bidirectional long short-term memory (Bi-LSTM) model and is enhanced with an attention mechanism. To validate the model's precision and efficiency, we compare its prediction results with the results of traditional full-wave electromagnetic simulations. The model demonstrates outstanding crosstalk prediction capabilities, achieving an accuracy exceeding 95.72%. Notably, the proposed method diminishes the full board prediction time from hours to mere seconds. To verify the model's reliability, the trained model was used to predict the crosstalk of three unseen PCBs and accurately identified the potential crosstalk in the unseen PCB. Thus, the proposed method can practically be applied to a fully automated and intelligent electromagnetic compatibility design of electronic products.