1 - 4 of 4
-
PublicationGenerative Machine Learning for Resource-Aware 5G and IoT SystemsExtrapolations predict that the sheer number of Internet-of-Things (IoT) devices will exceed 40 billion in the next five years. Hand-crafting specialized energy models and monitoring sub-systems for each type of device is error prone, costly, and sometimes infeasible. In order to detect abnormal or faulty behavior as well as inefficient resource usage autonomously, it is of tremendous importance to endow upcoming IoT and 5G devices with sufficient intelligence to deduce an energy model from their own resource usage data. Such models can in-turn be applied to predict upcoming resource consumption and to detect system behavior that deviates from normal states. To this end, we investigate a special class of undirected probabilistic graphical model, the so-called integer Markov random fields (IntMRF). On the one hand, this model learns a full generative probability distribution over all possible states of the system-allowing us to predict system states and to measure the probability of observed states. On the other hand, IntMRFs are themselves designed to consume as less resources as possible-e.g., faithful modelling of systems with an exponentially large number of states, by using only 8-bit unsigned integer arithmetic and less than 16KB memory. We explain how IntMRFs can be applied to model the resource consumption and the system behavior of an IoT device and a 5G core network component, both under various workloads. Our results suggest, that the machine learning model can represent important characteristics of our two test systems and deliver reasonable predictions of the power consumption.
-
PublicationEnergy harvesting and conversion - applications of piezoelectric transformer and transducer MEMS( 2018)A systematic investigation of the feasibility to integrate complete piezo-based power supply on silicon was done. Up to now, fully integrated off-line power supplies on chip are available as products for below 1 Watts. Higher power levels up to 10 Watts and more are strongly desired for many miniaturized applications as Off-Line LED light sources, integrated power supplies for communication devices as iPhone, portable devices for medical applications, portable beamers an others. The integration of high-efficient power supplies based on magnetic transformers (PT) including galvanic isolation is limited due to the physics of electromagnetism. Piezoelectric transformers can be integrated as MEMS when PZT material is applied on silicon to a height of several Micrometers to form an oscillating device which will be processed after micro-bonding in an etching process. Although power density of discrete PT is already high, it can be increased by a factor of 100 to 1000 in integrated devices on silicon taking advantage of uniform crystal structure of sputtering process and improved heat removal through silicon. Serial piezo-transformer-strings allow for high isolating voltage up to 4 kV and provide efficiency up to 95% or more, but unfortunately on the cost of significant large chip area. However, piezoelectric transformers will gain higher acceptance in power converters if a magnetic-field-free environment is requested as for magnetic resonance tomography. Promising piezoelectric applications can be found for transformer-transducer units to harvest ultrasonic energy, preferably in medical therapy-diagnosis applications, but further, in industrial autonomous sensor supplies with avoidance of electromagnetic disturbance. Piezoelectric energy harvesting becomes attractive using ultrasonic energy harvester MEMS with wide range frequency excitation using permanent magnet cantilever construction. Ultrasonic MEMS loudspeakers are miniaturized alternatives to traditional magnetic devices. The advantage of piezoelectric MEMS applications will result in an extreme miniaturization compared to conventional power conversion by magnetic or electrostatic solutions. High reliability including intelligent integrated functions in some cases may improve the practicability of piezoelectric MEMS.
-
PublicationMiniaturization of power converters by piezoelectric transformers - chances and challenges( 2017)A systematic approach of the feasibility to integrate complete piezo-based power supply on silicon is the focus of research activities within Fraunhofer EAS, ISIT an IZM. Up to now, fully integrated off-line power supplies on chip are available for below 1 Watts, e.g. from Texas Instruments. Higher power levels up to 10 Watts and more are strongly desired for many miniaturized applications as Off-Line LED light sources, integrated power supplies for communication devices as iPhone, portable devices for medical applications, portable beamers an others. The integration of high-efficient power supplies based on magnetic transformers (PT) including galvanic isolation is limited due to the physics of electromagnetism. Piezoelectric transformers can be integrated when PZT material is applied on silicon to a height of several Micrometers to form an oscillating device which will be processed after micro-bonding in an etching process. Although power density of discrete PT is already high, it can be increased by a factor of 100 to 1000 in integrated devices on silicon taking advantage of uniform crystal structure of sputtering process and improved heat removal through silicon. The driving topology can be formed by high-voltage Mosfets or multi-level low-voltage Mosfet topology based on SOI or GaN on Si and integrated micro-inductors in the future. Serial piezo-transformer-strings allow for high isolating voltage up to 4 kV and provide efficiency up to 95% or more. Synchronous rectifying devices can be formed by low-voltage Mosfets at the output stage of the power supply. The advantage will be an extreme miniaturization compared to discrete power supplies, reduction of blocking capacitors by interleaving techniques, and thus, high reliability including intelligent integrated functions as stabilization circuits, sensors or control.
-
PublicationEntwurfsmethoden für verbesserte robuste Batteriemanagementsysteme. Teilvorhaben(Fraunhofer IIS / EAS, 2016)
