Search

A 7.5 Gb/s/pin 8-Gb LPDDR5 SDRAM is implemented in a 1 $\times$ nm DRAM process. Various techniques are applied to achieve higher bandwidth and lower power than LPDDR4X. To increase data rate, a WCK clocking scheme that is less vulnerable to power noise is adopted and a non-target ODT mode is proposed to reduce reflection noise in a two-rank system. A couple of techniques are proposed for saving power. To reduce self-refresh power, this chip supports deep sleep mode (DSM). In DSM, the leakage current of internal voltages decreases by disabling internal voltage generators that are not related with a self-refresh operation. Dynamic voltage frequency scaling (DVFS) is adopted to reduce read and write operation power and when writing all zeros data, an internal data copy function can be used for reducing write operation power. Last, a ZQ calibration scheme that shares one ZQ resistor (RZQ) and automatically executes ZQ calibration is presented. The proposed LPDDR5 DRAM operates up to 7.5 Gb/s on an automatic test equipment (ATE) and 6.4 Gb/s on a prototype system. Read and write power decrease by 21% and 33% compared to LPDDR4X at 4.266 Gb/s, and self-refresh power is reduced by 25% in DSM.

This paper presents a deep neural network based test vector generation method for dynamic verification of memory devices. The proposed method is built on reinforcement learning framework, where the action is input stimulus on device pins and the reward is coverage score of target circuitry. The developed agent efficiently explores high-dimensional and large action space by using policy gradient method with A-nearest neighbor search, transfer learning, and replay buffer. The generated test vectors attained the coverage score of 100% for fifteen representative circuit blocks of modern DRAM design. The output vector length was only 7% of the human-created vector length.

The graphic DRAM standard GDDR6 is developed to overcome the limitation of previous standards GDDR5/5X for achieving high-speed operation. This paper introduces 16-Gb GDDR6 DRAM with a per-bit trainable single-ended decision feedback equalizer (DFE), a reference impedance (ZQ)-coded transmitter, and a phase-locked loop (PLL)-less clocking to overcome I/O speed limitation by the DRAM process. Furthermore, this paper optimizes clock- and power-domain crossings and adopts split-die architecture to improve signal integrity (SI). This GDDR6 operates 16 Gb/s/pin with 1.15 V and achieves 18 Gb/s/pin with 1.35 V in the DRAM process.

This paper presents a dual-loop two-step ZQ calibration scheme with a 20-nm DRAM process to support dedicated supply voltages ( $V_{DD}$ and $V_{DDQ}$ ). The proposed calibration scheme improves system signal integrity by maintaining the targeted output level ( $V_{\mathrm {OH}}$ ) in dynamic voltage–frequency scaling (DVFS) function, where $V_{DD}$ and $V_{DDQ}$ can be independently controlled to save system power. A two-step conversion algorithm alleviates the increase in calibration time, which is caused by an additional on-die termination (ODT) calibration for command/address (CA). The offset of a dynamic comparator in a ZQ calibration engine is averaged by a fraction-referred input switching-then-averaging (FISA) scheme which minimizes the effect of kickback noise. A code-referred periodic ZQ update (CPZU) scheme tracks the voltage and temperature variation during the background mode while minimizing the chance of malfunction by interference of both DRAM internal noise and external power noise. By applying the DVFS function based on the device usage by the user scenario in this design ( $V_{DD} =1.1$ V and $V_{DDQ} = 0.95$ V) at 1.6 Gbps, the read and the write operating power consumption can be saved by 17.7% and 18.2%, respectively, without a dynamic drop in performance. Therefore, the proposed dual-loop two-step ZQ calibration design improves the overall DRAM power efficiency based on the existing LPDDR4 SDRAM architecture without circuit overhead.

High-speed and low-power techniques for the latest mobile DRAMs, such as LPDDR4/4X [1–3], have been developed to enable high-resolution displays, multiple cameras and 4G communication in mobile devices. However, DRAM with higher bandwidth and lower power consumption than LPDDR4X is indispensable to support 5G communication, on-device artificial intelligence and advanced driver assistance systems. In this paper, we present a 1st generation 10nm-class process LPDDR5. It includes novel schemes that increase the maximum bandwidth, such as WCK clocking and non-target ODT (NT-ODT). Power consumption is also reduced by low power schemes, such as dynamic voltage-frequency scaling (DVFS), an internal data-copy function called write-X and a deep-sleep mode (DSM).

This paper proposes a degree of fault isolability concept and active fault diagnosis method for redundantly actuated vehicle systems. Fault isolability is a structural property related to system dynamics and composition of actuators and sensors. Existing research on testing fault isolability has involved checking whether the system is isolable, i.e., binary in nature. A continuous value rather than a binary metric is needed to evaluate how isolable a given system fault is based on a specific measurement set. After fault components are isolated, the fault type and magnitude are estimated by analyzing residual vectors. In a redundantly actuated system, the number of controls/actuators is greater than the system mobility. Thus, the control input distribution to achieve a given control objective is not unique. In the case of a fault, the active fault diagnosis system adjusts the control input distribution to diagnose the fault. Thus, much more system information can be identified by additional excitation through a redundantly actuated system, which improves the fault diagnosis performance. Simulation results of a four-wheel independently driven and steered vehicle model validated the proposed degree of fault isolability and the effectiveness of the proposed active fault diagnosis method.

Starting at 512Mb 6Gb/s/pin [1], GDDR5's speed and density have been steadily developing for about 10 years; recently achieving 8Gb 9Gb/s/pin [2] with per-pin timing training. Although 8Gb GDDR5X can operate at 12Gb/s [3] by increasing the burst length (BL) from 8 to 16, a degradation in system performance at a data granularity of 64B is seen. The I/O specification, using PLL clocking that additionally causes PLL jitter, has not changed much compared with GDDR5. To overcome these issues, GDDR6 introduced a dual channel for a data granularity of 32B with a BL16, per-bit training of l/ REF , and an equalizer with PLL-less clocking. This paper presents a 16Gb 18Gb/s/pin GDDR6 DRAM with a die architecture and high-speed circuit techniques on 1.35V DRAM process.

This paper presents a dual-loop 2-step ZQ calibration scheme with 20nm DRAM process to support dedicated supply voltage (VDD, VDDQ). The proposed calibration scheme maintains a target value of on-die termination (ODT) in DQ/CA regardless of the supply-voltage variations which are caused by dynamic voltage frequency switching (DVFS) and alleviates the calibration time which is increased by insertion of additional CA calibration. The offset of a comparator is averaged out by fraction-referred input switching-then-averaging scheme (FISA). And code-referred periodic ZQ update (CPZU) scheme can track the VT variation while minimizing the interference.

With growing demand for low-power mobile applications, such as wearable devices, smart phones and tablet PCs, low-power mobile DRAM has been identified as a mandatory requirement for low-power system designs. The recently developed LPDDR4 [1] is still a power efficient solution because of its architectural approaches and low-voltage-swing terminated logic (LVSTL). However, demand for enhanced power-efficiency beyond LPDDR4 is still increasing for mobile applications. In this work, a 5.0Gbp/s/pin 8Gb LPDDR4X memory with power-isolated low-voltage-swing terminated logic (PI-LVSTL) and a split-die architecture is proposed to enhance power-efficiency and mass production yield.

Finite element model updating is an optimization problem to identify and correct uncertain modeling parameters. In conventional model updating, physically incompatible criteria, which designate differences between analytical and experimental results, are combined into a single-objective function using weighting factors. There are no general rules for selecting the weighting factors since they are not directly related to the dynamic behavior of the updated model. Thus, a necessary approach is to solve the time-consuming optimization problem repeatedly by varying the values of weighting factors until a satisfactory solution is obtained. In this work, an interactive multiobjective optimization technique called satisficing trade-off method is introduced to avoid the difficulty. It is relatively easy to state what kind of solutions is satisfactory considering the correlations of the initial FE model with the experimental results, the importance of individual modal properties, and measurement errors, etc. The satisficing trade-off method uses this information directly in the optimization process and finds a Pareto solution which is nearest to the given information. Moreover, as the method provides the tangent hyperplane which approximates the Pareto surface in the neighborhood of the obtained Pareto solution, the desired updated model can be found in a few iterations.

This paper describes techniques used to analyze the sweet spot of sound field reproduced by ear-level linear arrays of loudspeakers by geometrical approach method. Previous researches have introduced various sweet spot definitions in their own way. In general, sweet spot is defined as an area whose stereophonic sound effect is valid. Its size is affected by the geometrical arrangement of the system. In this paper, a case when plane waves are generated by linear arrays of loudspeakers in the horizontal plane is considered. So the sweet spot is defined as an area in which the listener can perceive the desired azimuth angle. Because there are many loudspeakers, impulse responses at listener`s ears are in the form of pulse-train and the time-duration of the pulse-train affects the localization performance of the listener. So we calculated the maximum time duration of pulse-train by geometrical approach method and identified with the results of impulse response simulation. This paper also includes parameter analysis with respect to aperture size, so it suggests a tool for sound engineers to expect the sweet spot size and listener`s sound perception.

Observable mathematical model of DDM (Direct Dirve Motor) was suggested. The motor that operates the object system directly is called DDM. DDM has many strong points, however, it has a significant disadvantage, that it is more sensitive to the external force than the motor with reduction gear. In other word, if the force is applied, air gap of the motor can be perturbed. This causes not only difficulty in motor control but also even more serious problem, such as the breakdown of motor. However, if the air gap variation can be estimated, it can help prevent these problems. DDM should be modeled to estimate the air gap variation. The type of researched DDM is PMSM (Permanent Magnet Synchronous Motor) and precedent model of PMSM includes only characteristics of electro-magnetic system and rotational motion. However, suggested model should also include characteristics of translational motion of rotor to estimate the air gap variation. Also, this model should satisfy observability condition, because state observer is designed based on this model.

A novel updating method of equivalent factor of the Equivalent Consumption Minimization Strategy (ECMS) is proposed to prolong the lifetime of battery in fuel cell hybrid electric vehicle. ECMS can generate the sub-optimal solution at any instance for current driving condition. Equivalent factor which is dimensionless conversion ratio of electrical power flow into chemical power flow plays an important role in charge-sustaining performance. Well-tuned equivalent factor from iterative search in off-line can only guarantee the charge-sustaining performance, not the lifetime of the battery. Therefore additional updating method of equivalent factor is necessary in order to prolong the lifetime of the battery which only takes the numbers of charge and discharge cycle into account. Backward simulation is used to verify the effectiveness of the updating method.

A variety of noise control methods have been developed as an interest on noise issues increases. Among them, noise control methods using masking effect, a phenomenon to reduce the ability to notice the unwanted sound by proper sound, to implement a pleasant sound environment have been studied under the name of soundscape. We proposed a novel vector base amplitude panning(VBAP) based noise control method to apply to the building environment. The proposed method could improve the amenity inside the building to reproduce the sounds with excellent masking effect on the incoming path of noise using the control speakers, considering the direction of noise source. The directional masking sounds can be generated by using VBPA technique. To verify the performance of the proposed method, we carried out the subjective test for the degree of amenity according to direction of the masking sound. Subjective test results showed that it is possible to improve the amenity inside the building by controlling the direction of masking sound considering the human`s auditory characteristic.

Tilt sensor is required to control the attitude of the biped robot when it walks on an uneven terrain. Vision sensor, which is used for recognizing human or detecting obstacles, can be used as a tilt angle sensor by comparing current image and reference image. However, vision sensor alone has a lot of technological limitations to control biped robot such as low sampling frequency and estimation time delay. In order to verify limitations of vision sensor, experimental setup of an inverted pendulum, which represents pitch motion of the walking or running robot, is used and it is proved that only vision sensor cannot control an inverted pendulum mainly because of the time delay. In this paper, to overcome limitations of vision sensor, Kalman filter for the multi-rate sensor fusion algorithm is applied with low-quality gyro sensor. It solves limitations of the vision sensor as well as eliminates drift of gyro sensor. Through the experiment of an inverted pendulum control, it is found that the tilt estimation performance of fusion sensor is greatly improved enough to control the attitude of an inverted pendulum.

We propose a novel design of an artificial robot ear for sound direction estimation using two measured outputs only. The spectral features in the interaural transfer functions (ITFs) of the proposed artificial ears are distinctive and move monotonically according to the sound direction. Thus, these features provide effective sound cues to estimate sound direction using the measured two output signals. Bilateral asymmetry of microphone positions can enhance the estimation performance even in the median plane where interaural differences vanish. We propose a localization method to estimate the lateral and vertical angles simultaneously. The lateral angle is estimated using interaural time difference and Woodworth and Schlosberg's formula, and the front-back discrimination is achieved by finding the spectral features in the ITF estimated from two measured outputs. The vertical angle of a sound source in the frontal region is estimated by comparing the spectral features in the estimated ITF with those in the database built in an anechoic chamber. The feasibility of the designed artificial ear and the estimation method were verified in a real environment. In the experiment, it was shown that both the front-back discrimination and the sound direction estimation in the frontal region can be achieved with reasonable accuracy. Thus, we expect that robots with the proposed artificial ear can estimate the direction of speaker from two output signals only.

A variety of methods for sound source localization have been developed and applied to several applications such as noise detection system, surveillance system, teleconference system, robot auditory system and so on. In the previous work, we proposed the sound source localization using the spatially mapped GCC functions based on TDOA for robot auditory system. Performance of the proposed one for the noise effect and estimation resolution was verified with the real environmental experiment under the single source assumption. However, since multi-talker case is general in human-robot interaction, multiple source localization approaches are necessary. In this paper, the proposed localization method under the single source assumption is modified to be suitable for multiple source localization. When there are two sources which are correlated, the spatially mapped GCC function for localization has three peaks at the real source locations and imaginary source location. However if two sources are uncorrelated, that has only two peaks at the real source positions. Using these characteristics, we modify the proposed localization method for the multiple source cases. Experiments with human speeches in the real environment are carried out to evaluate the performance of the proposed method for multiple source localization. In the experiments, mean value of estimation error is about and percentage of multiple source localization is about 62% on average.

When we render 3D sound images using headphones or speakers, the main key of this technology is the Head-related transfer function (HRTF) database. Even though there are various HRTF databases, they have some drawbacks such as detrimental effects caused by imperfect measuring environment and insufficient measurement points. Moreover there is no database with Korean subjects. We are planning to develop the HRTF database for Korean. As a first step to establish the HRTF database aimed at Korean, the new HRTF measurement system with minimized aforementioned drawbacks is designed. In this paper, the newly designed HRTF measurement system is introduced and the overall effects caused by the diffraction of the apparatus, especially the headrest and backrest of the chair, are analyzed. The backrest of the chair does not distort the HRTFs significantly while the headrest makes significant distortion on the HRTFs and it could have significant effects on directional perception. We determined acceptable head rotation angle and head position of the subject for accurate HRTF measurement based on the experiments with B&K HATS. We conclude that the 3 degrees of the head rotation and the 1.5cm front/back/left/right shift of the head do not distort the HRTFs significantly.

This study deals with the perceptual sensitivity to Head-Related Impulse Responses (HRIRs) in the median plane based on a series of subjective listening tests using a pair of headphones. First, the non-individualized HRIRs were modeled from 12 principal components (PCs) extracted from Principal Components Analysis (PCA) of the CIPIC HRTF database. The Just Noticeable Difference (JND) in weight of PCs (PCWs) at each elevation was estimated. It was not observed the common elevation-dependent tendency or PCW-dependent tendency of JND in PCWs across the five subjects who participated in the tests, and the inter-subject variation of JND in PCWs was large. The JND in HRIRs can be estimated indirectly from the JND in PCWs because the HRIRs can be represented by a linear summation of the PCs weighted by PCWs. The common elevation-dependent tendency of JND in Directional Impulse Responses (DIRs), which are the mean-subtracted HRIRs, across the five subjects can be found. The change in PCWs does not seem to contribute to our perception of sound source characteristics; however, the resulting change in HRIRs due to the change in PCWs seems to contribute. The subjects showed larger JND in DIRs in the frontal region than in the rear region. This means that our perception of sound source characteristics is more sensitive for frontal sources than rear sources.

We propose a novel probabilistic method for quantitative analysis of the sound localization performance. The analysis is based on the two kinds of probability distributions estimated from a single dataset containing listening test results for sound localization. The quantitative parameters of the von Mises probability distributions provide meaningful interpretations on the localization performance. The mean direction represents the listener’s perceptual bias, and the shape parameters and the circular variance provide information on how much the responses are concentrated about the mean direction. The front–back confusion can be determined more systematically by the proposed method than the conventional one, especially for the responses near the boundary of front–back confusion region based on the conventional criterion. The proposed method can be easily extended to analyze the up–down and left–right confusions. To investigate the feasibility of the proposed method, the already published dataset originally obtained by Iida et al. was analyzed using the proposed probabilistic method. The results showed that the proposed method can provide meaningful and reasonable interpretations on the localization performance.

Sound source localization method based on the time delay of arrival(TDOA) is applied to many research fields such as a robot auditory system, teleconferencing and so on. When multi-microphones are utilized to localize the source in 3 dimensional space, the conventional localization methods based on TDOA decide the actual source position using the TDOAs from all microphone arrays and the detection measure, which represents the errors between the actual source position and the estimated ones. Performance of these methods usually depends on the number of microphones because it determines the resolution of an estimated position. In this paper, we proposed the localization method using spatially mapped GCC functions. The proposed method does not use just TDOA for localization such as previous ones but it uses spatially mapped GCC functions which is the cross correlation function mapped by an appropriate mapping function over the spatial coordinate. A number of the spatially mapped GCC functions are summed to a single function over the global coordinate and then the actual source position is determined based on the summed GCC function. Performance of the proposed method for the noise effect and estimation resolution is verified with the real environmental experiment. The mean value of estimation error of the proposed method is much smaller than the one based on the conventional ones and the percentage of correct estimation is improved by 30% when the error bound is .

Preview control involves using forward road information acquired from preview sensors in designing a controller. Preview control, which is composed of feedback and feed-forward input parts can lead to better performance than feedback control alone. In this paper, application of the preview control based on the active suspension to tracked vehicles will be introduced. The suspension unit of the tracked vehicle is in the form of a trailing arm type to increase the wheel travel. Because the preview control has been developed for the linear time-invariant system, a linearized model of the system is required. A linear model of the tracked vehicle, which the preview controller is designed based upon, is generated without considering the dynamics of the track elements for simplification. On the other hand, the nonlinear model, which the performance of the controlled system is investigated based upon, is generated by using Recurdyn™, commercial software. The nonlinear model includes nonlinear damping and stiffness characteristics of hydraulic suspension units(HSUs). From the simulation results for RRC 9, performance enhancement of the preview control will be shown in terms of ride comfort by using ISO 2631 under the assumption that actuators are ideal and the velocity of the vehicle is constant.

New approximate models for external acoustics interacting with flexible structures are developed. The basic form of the present models is obtained by a combination of the Laplace-transformed retarded and advanced potentials with the weighting parameter as part of the model equation. It is shown that the maximum attainable time-derivative of convergent approximate models is two, hence any attempt to include higher orders will lead to non-convergent models. The present external acoustic model is implemented and interfaced with a finite element structural analyzer for the transient response analysis of submerged spherical and cylindrical shells subjected to a series of incident waves. Comparisons of the present results with the classical analytical solutions and the Doubly Asymptotic Approximations (DAAs) show that proposed model offers improved accuracy especially for early-time responses, exhibits computational robustness, and maintains the impulse response consistency that are desirable for inverse problems. The present model is applied for a shock response correlation of an experimental complex ring cylinder, which demonstrates the applicability of the proposed approximate model to practical engineering problems.