The total-ionizing-dose response of partially-depleted silicon-on-insulator devices with or without grounded substrate under different irradiation bias is investigated. Compared with devices with grounded substrate, the devices with floating substrate introduces more trapped positive charges in buried oxide after irradiation, leading to larger negative shift of back-gate threshold voltage, especially for ON bias condition. Theoretical analysis and TCAD simulation demonstrate that, this phenomenon is attributed to the increased initial built-in field in buried oxide and weakened space charge effect under the case of floating substrate. According to this work, substrate terminal of SOI devices must be considered for integrated circuit design under radiation condition.
This letter proposes a new state estimator called the multiple-model hybrid particle/finite-impulse-response (FIR) filter (MMHPFF) for indoor localization using wireless sensor networks. In the proposed hybrid filtering algorithm, the multiple-model particle filter has the role of the main filter, and it overcomes uncertain process noise problems arising from the use of the constant velocity (CV) motion model in indoor localization. In addition, the multiple-model FIR filter is used as an assisting filter to overcome particle filter failures owing to the sample impoverishment phenomenon. Indoor localization simulations demonstrated that the proposed MMHPFF is more accurate and reliable than conventional algorithms.
In this paper, a low critical path delay (CPD) circuit structure is proposed for composite field S-box circuit. In the low CPD structure, multiplicative inverse over GF(24) and multiplicative over GF(24) are constructed by AND-XOR-networks. The XOR-networks in the last two multiplications over GF(24) are further merged with the following constant matrix multiplication operation to shorten the CPD. Finally, hardware complexities of our designs are compared with previous works. The comparisons indicate that our proposed method is effective. Our design of S-box/InvS-box based on the proposed method has lower CPD.
A low-power Colpitts VCO with high output efficiency that uses base inductive feedback and Q-factor enhancement techniques is proposed in this paper. The base inductive feedback technique employs an inductor at the base of the bipolar transistor to generate a feedback enhancement effect and reduce the start-up current. The Q-factor enhancement technique adopts a capacitor voltage divider at the output of the Colpitts VCO to improve the DC-to-RF efficiency. The mechanisms are theoretically analyzed, and then a 433 MHz Colpitts VCO is designed to verify the scheme. Its DC consumption is as low as 270 µW while the operating frequency is 433 MHz. Finally, the Colpitts VCO is applied in a wireless neural signal recording system and works well. Thus, the presented VCO is suitable for analog signal acquisition system in the extremely power-constrained wireless scenarios.
Approximate Network on Chip (NoC) is being considered as a good solution to reduce power consumption and improve the communication efficiency. However, increasing Multi-Cell Upsets (MCU) is seriously threatening the approximate NoC reliability and causes output far away from guaranteed quality of results. In this paper, we propose an unequal message protection framework with same coding length for mitigating MCU impacts on approximate NoC. The proposed framework makes good use of side information to select the special messages from normal messages and embeds same length coding scheme for all messages while providing stronger protection capability for special messages in approximate NoCs. The effectiveness of proposed methodology is evaluated by complete fault injection, which captures the dual-voltage approximate technique, unequal message protection mechanism and MCU error propagation in a joint manner. The cycle-accurate simulation using synthetic (e.g., Transpose, Uniform) and realistic benchmarks (e.g., VOPD and MWD) show that, compared with the state-of-the-art approximate NoC architecture, the proposed unequal message protection framework can achieve up to average 84.9% reliability improvement with 6.25% power consumption 6.9% area overhead, and 15.38% latency increase.
This letter presents an injection-locked 36 GHz frequency tripler (ILFT) for 5G applications. With the high order transformer based LC tank, over the target operating frequency range its phase response is flat at around 0 degree, and the injection-current is enhanced with a 3rd harmonic LC tail filter, thereby achieving a wide locking range. Fabricated in a 65 nm CMOS process, the ILFT consumes about 7.2 mW from a 0.6 V supply, and its core area without the output buffer and input matching network is only 0.25 × 0.4 mm2. With 0 dBm input injection power, the proposed ILFT achieves a locking range from 31.5 to 40.5 GHz (25%).
A novel VCO auto frequency calibration (AFC) technique is presented in this paper. It provides ultra-fast and high-precision search for an optimum VCO discrete tuning curve among a group of frequency sub-bands in the PLL frequency synthesizer. The calibration circuit consists of a table of frequency (TOF) and a finite state machine (FSM). Asynchronous work of TOF and FSM makes it possible to realize ultra-fast and high-precision calibration simultaneously. TOF affords the high-precision of the calibration and FSM with a high-speed clock offers the ultra-fast search. A 2.8 G–3.5 GHz Fractional-N PLL incorporating the proposed calibration technique is implemented in 55 nm CMOS process. The measured calibration time for a 5-bit capacitor bank is 90 ns for a frequency resolution of 203 KHz. Such small calibration time significantly reduces the effects of calibration time on the total lock time of the PLL.
Compare with circular coils, Double-D coils are widely used in the wireless charging system of electric vehicles for its high coupling coefficient and excellent offset tolerance. In order to obtain higher coupling coefficient and reduce the leakage magnetic flux in the central part of the coils, this paper proposes a method that optimizing stray parameters, which can achieve more than 95% of transmitting efficiency. The investigation is performed by joint simulation for the choice of experimental devices concerning current-withstanding characteristic. According to simulations and experiments, more than 11 kW output power can be achieved for a wide range of frequency.