Search

The control of quality is an obligatory stage of any welding work which makes problem of timely detection of welding defects is important for industry nowadays. Through digitisation it possible to solve this problem more efficiently and eliminate the human factor. This paper presents a real-time industrial automated video analytics system for weld defect detection. The system uses deep learning to analyze video images and detect weld defects in real-time. Experiments showed that the system can effectively detect weld defects with high accuracy and speed. A safety and trustworthiness analysis of the system showed that it can be reliable and safe for use in industrial manufacturing. The proposed solution has several key advantages for the industry: it allows for real-time data processinga and requires low computing power, making it an energy-efficient and cost-effective solution.

Quality of life largely depends on the ability of a person to resist stress, what refers to the concept of resilience to stress. The priority groups are older and frail people, disabled people and people under strain conditions. The autonomic resilience, along with the motor and cognitive resilience, is regarded as one of the major components of resilience. In frail people, autonomic resilience is substantially reduced. In this concept position paper, we explore human autonomic sensorics when an mHealth system provides digital assistance in autonomic functioning of a person during her/his daily activity. The core of our concept is a smartphone which can used in person’s daily activity for monitoring such critical markers of the human autonomic function as orthostatic hypotension during transition from supine or sitting to standing position. A smartphone has potential to sense both blood pressure and heart rate during events of postural transition. The developed mHealth concept aims at digital support of the autonomic activity of a person in his/her daily life. The provided information encourages the person for better performance and results.

In the modern world, tourists are faced with the need to use various mobile applications for travel planning and navigation. To provide tourists with comprehensive information about the visited area, a digital model can be constructed, so implementing a digital shadow of given territory and related processes. In this paper, we discuss our concept of a platform for smart e-Tourism services. Such services provide tourists with various information about geographical locations and points of interests as well as about related facts, persons, and stories. A high-level architecture for the services is presented. Enabler algorithms are considered for implementation of the services. The proposed concept support tourists with latest information about the properties of the physical environment, news, social networks, and the number of visitors in relation to geographical locations.

Video Surveillance Systems (VSS) are progressing towards the industry digitalization. In this demo, we apply the Artificial Intelligence (AI) technology to detection of key points for a rainbow trout. The well-known YOLO-Pose neural network is typically used for human pose recognition in video recordings. We use this neural network for detection of 8 key point for an individual fish. Our constructed dataset includes 10 500 images for approximately 30 thousand fishes. The YOLO-Pose network is trained on the dataset using our combined loss function. Our early experiment shows that that detection works fine even in muddy underwater conditions.

The progress in mobile technologies of Internet of Things (IoT) effectively supports the digitalization of sports training equipment. In this demo, we show an assistant mobile application (digital arbiter) to track the quality of exercise performance by the user. We experiment with sports training equipment produced by the MB Barbell company. The exercise is bench press from the chest or similar. The equipment has a mechanism to attach user’s smartphone. The problem is to estimate the lower and upper bounds for each round of execution. The measurement is based on data coming in real-time from the smartphone accelerometer. First, the mobile application can check the correctness as reaching the bounds. Second, the distance and time are estimated that the user has overcome during the exercise rounds. Since each user has own characteristics (e.g., arm length, grip method), individual calibration of the sensor system is needed before exercise performance on a given sports training equipment. Our experiments show that the mobile application has good accuracy and real-time performance, so the digital arbiter can be used as in individual daily physical activity or in mass sports events.

The progress in technologies of Internet of Things (IoT) and Artificial Intelligence (AI) lead to the digitalization in various domains of human activity. In this demo, we show a digital inspector that recognizes the physical human activity with sports training equipment (mechanical simulators). We experiment with outdoor simulators produced by the MB Barbell company and located on the shore of Lake Onega in the Petrozavodsk city for free use by citizens and guests. We implemented multi-camera video surveillance system (VSS) that statistically evaluates the states of the sports equipment use such as “the simulator is occupied by a person”, “the simulator is not correctly used”, “the simulator is occupied but not in use”, etc. The evaluation is based on well-known AI methods for image recognition, which are customized for real-time inspection. The evaluated statistics for the area with sports equipment is important for the municipality to understand the efficiency of the provided public resources for well-being and comfort life.

The digitalization of sport training machines enables sensor-based applications for recognition of human movement at exercise performing. In this demo, we continue our development of the mobile application that uses evaluation of athlete's hands position in real-time. We show more effective solution (in terms of the position accuracy) than we demonstrated at the previous FRUCT conferences. Our previous solution is based on an accelerometer as a sensor for input data. Our successor solution combines an accelerometer and a gyroscope based on Kalman filter. This combination reduces the influence of acceleration on evaluating the angle of the lever of sport training machine relative to the vertical axis (the ``Bench Press'' exercise is used as a demo use case). The accurate measurement of hands position supports estimation of the total distance passed by hands (with given weight). This metric is important for training as well as for new class of sport competitions.

The Quality of Life (QoL), especially of frail and older people, depends on the ability of a person to tolerate stresses and its negative consequences. The toleration is related to the human resilience, reflecting the three basic human functions: cognitive (the thinking process), motor (the movement activity), and autonomous (the internal body processes). The cognitive resilience is acknowledged as one of the main components of human resilience. The cognitive resilience is tightly coupled with the motor resilience. Various motor-cognitive tests are known to assess the reactivity in a custom way and with specialized medical equipment. In this paper, we introduce our concept model of the mobile Health (mHealth) sensorics for digital assistance of the human resilience during person’s daily activity. In reactivity testing, a smartphone becomes a core element, replacing specialized medical equipment and transforming to a daily automatic instrument of human motor-cognitive sensorics. The collected measurements provide the base to monitor person’s mental condition.

Digitalized evaluation of human gait becomes a topical problem for mobile healthcare (mHealth) and well-being. We focus on the evaluation for a person performing diagnostic physical exercises. We introduce our algorithm for constructing a digital model of human gait when the person is performing a given exercise. The presented demo application observes human gait based on sensed data from Virtual Reality (VR) trackers attached to human body. Sensed data are collected in the database and processed to construct a 3D model of human motion based on a mathematical model for estimating the human gait parameters (metrics). The VR technology is used for visualization both when a person executes the given exercise and when constructing the resultant 3D model to analyze. Our experimental setup ensures the interconnection of the equipment to perform exercises in virtual space and organize the collection and processing of sensed data. VR trackers operates in real time. Based on the collected data, numerical motion metrics are estimated to form a digital model of human gait. The 3D model is then used by a doctor to analyze gait disorders or by the person him/herself.

Many multi-camera systems organize video surveillance - a process designed to visually control a certain area and / or objects. The most modern systems organize smart video surveillance - the process of collecting and subsequent analysis of incoming information using modern technologies and designed to ensure security in those places where it is necessary. At the moment, there are a large number of algorithms and methods used in video analytics systems - a technology that uses computer vision methods for automated data acquisition based on the analysis of images from video cameras in real time. Such video analytics systems are mainly used in special services. In the case of commercial and public use, they are expensive or not available. The claimed system can be quite flexibly upgraded to suit the customer’s tasks, while having a low cost. At the moment, security systems in the field of video analytics are not widely used in the Republic of Karelia (RK). Video analytics is either completely absent or presented in the form of the most primitive recognition algorithms. Ensuring the digital security of the Republic of Karelia in public places and crowded places and the development of digital security support technologies in the region are among the highest priority areas of this work.

Robotic manipulators are widely employed in industry for automating repetitive tasks. A particular use case is sorting of waste materials. Movement of such a manipulator is subject for control since defects can occur during the operation. The assessment of operation quality can be also delegated from human to machine based on the progress in Artificial Intelligence (AI). In this demo, we consider video-based assessment as a promising approach for industrial robotics. We demonstrate our computer vision-based procedure for assessing the operation quality of a robotic manipulator. The experimental setup uses a single camera to detect successful and unsuccessful manipulations with objects, including the object loss during transmission in real time.

Currently, most of the powerlifting competitions are conducted without any digital systems, which will be responsible for the fairness of evaluation of participants. Since the evaluation of participants is carried out by judges, the following problems follow: Errors when checking the correctness of the exercise A large number of judges are needed to conduct competitions Occurrence of errors when logging competitions (entering all results manually) To solve this problem, we have decided to create the prototype of a digital referee system designed to analyze the correctness of the athlete's performance of the Chest Press exercise on the MB Barbell Chest Press / Bench Press simulator, based on the readings of a three-axis accelerometer provided by MB Barbell. The developed prototype allows to fix the initial and final positions of the athlete's hands, with the calculation of the movement of the hands performing the exercises. Before performing the working approach, calibration is performed, consisting of an arbitrary number of repetitions on a simulator with an empty weight. Calibration determines the extreme positions of the hands when performing an exercise on this simulator for this athlete. All calibration measurement results are entered into a special calibration table, according to which, in the future, the average calibration value of the amplitude of the athlete's arm movement is calculated. Thus, the developed system has the following functions: Fixing the correctness of the exercise without the participation of a judge The function of individual calibration for each athlete Implementation of the electronic protocol of the competition in real time with integration on the electronic scoreboard

Recent progress in Ambient Intelligence (AmI) technology leads to many proposals of sensing systems in the physical, informational, and social worlds. Such systems focus primarily on the sensed data retrieval, including advanced cases of Internet of Things (IoT) sensors and video capture. Operation of a sensor is subject to faults, anomalies, and outliers. In this work-in-progress paper, we consider the problem of tracking the operation of a video camera or environment sensor to recognize deviations from the expected function. The recognition is based on neural network and context analysis. We implemented a module that makes a sensor smart in terms of operation status recognition and its self-understanding. Our demo shows the applicability of the module in IoT monitoring systems.

Many video surveillance systems (VSS) have been already developed for various application domains. Such systemsare based on well-elaborated recognition algorithms of Artificial Intelligence (AI) and implemented as Ambient Intelligence (AmI) services in Internet of Things (IoT) environments. In particular, algorithms support such smart VSS functions of video data processing as human detection, human identification, object location within an image, human activity recognition. Many software tools have been developed to implement various recognition algorithms for VSS development. In this paper, we consider the following VSS development problems: a) a generic model of events in video data for a given problem domain, b) a hardwaresoftware architecture for data processing with existing recognition algorithms, and c) a model to construct a required smart VSS function using existing software tools. We introduce our event-oriented approach to solve the above VSS development problems. The approach is experienced in several use cases. Our experimental study shows the applicability of the proposed approach in terms of the accuracy and performance.

The quality of life depends on the human resilience to stresses and other negative impacts from the environment and society. We assume that the human resilience can be effectively supported by motor activity. The problem is that people reduce the motor activity (the human mobility) although they are potentially able to move (the human motility). In this paper, we consider human motor activity sensorics and study the concept of mobile Health (mHealth) system to digitally support the mobility of a person during her/his daily life. The sensorics is based on inertial sensors of a smartphone that accompanies the person. The smartphone evaluates various motor tests for the human activity. The collected statistics provide an interesting picture to motivate the person to more activity. We introduce the concept model that interrelates human resilience and motor activity. We discuss possible digital support of human resilience based on testing the human activity. In sum, this study contributes our concept of smartphone-based mHealth system that digitally supports the motor activity of a person in daily life subject to increase the human resilience.

Modern video surveillance systems (VSS) can recognize a person and her/his activity. A particular example of such a VSS system is the monitoring and support of human health and wellbeing during exercise with sports equipment. In this work-inprogress paper, a prototype of the VSS system is demonstrated for monitoring the movement of a person around the sport training machine. For the recognition problem we consider the following situations: a) training machine is free (no person is close to the machine); b) a person is in the area near the training machine (but not in the working area); c) a person is in the working area of training machine (but the training machine is not used); d) a person is working on the training machine. Our early experimental study evaluates the feasibility of such a VSS system.

Mobile devices in Internet of Things (IoT) can accompany human (or robot) when moving on the area. Many e-tourism platforms exist to collect data on points of interests located in some exhibition area. Artificial Intelligence (AI) provides methods to derive knowledge on recognize human activity and to recognize persons interests in exhibition area space. In this paper, we propose to apply AI methods to IoT-enabled sensed data in respect to the observed human activity. The research goal is to elaborate possible smart extends to e-tourism platforms. Our case study is the CycleAdvisor system. We introduce a concept model of a smart tourism recommender system that uses human-accompanying IoT devices to sense the human activity. From the sensed data persons interests are recognized and then used for constructing recommendations.

This study presents the concept of a computationally efficient machine learning (ML) model for diagnosing and monitoring Parkinson’s disease (PD) using rest-state EEG signals (rs-EEG) from 20 PD subjects and 20 normal control (NC) subjects at a sampling rate of 128 Hz. Based on the comparative analysis of the effectiveness of entropy calculation methods, fuzzy entropy showed the best results in diagnosing and monitoring PD using rs-EEG, with classification accuracy (ARKF) of ~99.9%. The most important frequency range of rs-EEG for PD-based diagnostics lies in the range of 0–4 Hz, and the most informative signals were mainly received from the right hemisphere of the head. It was also found that ARKF significantly decreased as the length of rs-EEG segments decreased from 1000 to 150 samples. Using a procedure for selecting the most informative features, it was possible to reduce the computational costs of classification by 11 times, while maintaining an ARKF ~99.9%. The proposed method can be used in the healthcare internet of things (H-IoT), where low-performance edge devices can implement ML sensors to enhance human resilience to PD.

Positioning of a robot in its autonomous movement is applied to the automatic control and trajectory planning. There are two types of systems: sensed data come from physical parameters sensors and video cameras. In this paper, we consider the integration of both data sources, providing a novel way for trajectory construction. The sensed data are collected from the IMU-sensor, camera, and Lidar. We present an architecture for collecting data from many sources and for using the data to evaluate the position of a mobile robot. The results of our early experiments confirm the feasibility of the data integration when a robot moves between special points with different sequences of actions. We also present an algorithm for the movement of a mobile robot between given points without loss of positioning accuracy at long distances. The accuracy is not only affected the Euclidean coordinates, but the orientation angle is also improved. As a result, a robot can be used in narrow indoor, where a special point can be reached only at a certain orientation angle.

Internet content provides much information about a particular person, especially related to her/his activity reflected in social networks as web-pages or attached files. The problem is to find such information for a given person and for given search query constraints. A lot of data sources have to be analyzed to find information to construct digital persons profile. In this paper, we consider iterative Internet search algorithm that generates multiple search queries and evaluates the found links on the relation level to the given person. A rank is computed for each link that shows the relation level to the given person. Our experiments show that the method finds links among which 75% or more is essentially related to the person. When the content of a link is further analyzed (e.g., parsing) the search quality increases to 90% and more.

In this paper, we explore the applicability of CNN for the classification of bearing defects. The approaches to the CNN evaluation are discussed and the one that brings the formulation of the diagnostic problem closer to real conditions is purposed. We demonstrate problems associated with the practical application of CNN in monitoring industrial equipment. We show that hyperparameter optimization is able to improve training process stability and consequently provide the reliability of the CNN-based diagnostic method. Our early experiments indicate a possibility of application CNN for failure diagnostics using a vibration signal provided that the training data contains a sufficient number of bearings with various types of faults, ensuring high representativeness of the dataset.

Recent technologies of industrial monitoring provide highly fragmented information. Large amounts of multiparameter sensed data on production process and equipment operation are stored in disparate structures (databases). Effective use of the collected information requires data fusion within an information-driven monitoring system. In this short paper, we design a semantic data model to create a unified information space that fuses events derived from real-time sensed data streams. Our semantic data model considers a hierarchy of production equipment nodes and supports rules for identifying and composing events in the production process under monitoring.

The data sensing and communication technologies of Industrial Internet of Things (IIoT) enables monitoring technical state and utilization conditions for industrial rotary machinery. The monitoring system is based on multiple sensors that embed or surround machinery unit under the observation. The sensed data are used for diagnostics of the machinery operation and utilization processes. In this paper, we construct a digital profile for a given machinery. Its digital profile is constructed from the sensed data as information model evolving in time. Hence, detection of deviations in the digital profile provides basic information on possible faults and other incorrectness in industrial rotary machinery. The NASA bearing dataset was used to evaluate the proposed model efficiency.

The performance of multi-stream sensed data processing is a challenging problem for Industrial Internet of Things (IIoT) monitoring applications. This short paper discusses an edge computing model for receiving and processing the sensed data. The key element of our model is specialized computing modules for reading raw sensed data from multiple sensors in the physical environment. We experimentally evaluate the performance varying the number of attached sensors.

The paper describes the architecture and intellectual tasks of an intelligent robotic system for performing reforestation work. The robotic system consists of two devices: a robotic mobile device and a local control station. The robotic system can operate independently and in semi-autonomous mode under the control of the operator.

Autonomous mobile robots have been discovering recently a wide range of applications in various areas of human activities. A robot uses many data sources (sensors) to recognize its current situation, including video, inertial, acoustic, mechanical stress. The data flows from such sources are redundant, error prone, delivered with a rather high rate, and contain considerable information as for the ongoing events so present measurement noise, errors and insignificant data about minor fluctuations of the environment and the device. They present a source information for the inference modules such as navigation, localization, path planning, scheduling etc. The total number of the ongoing events can be large. They have different importance for the movement control. In this paper, we consider a method for intelligent data selection. The method is based on the Additive Increase Multiplicative Decrease (AIMD) algorithm. The effect is in filtering the most significant information to forward to the robot movement control.

Industrial Internet of Things (IIoT) supports multiparameter monitoring systems, where much data are continuously sensed and processed on the edge in real-time. In this paper, we consider the service development problem for such IIoT data. We study the opportunities of human-computer interface that make such services smarter: a) mobile assistance, b) interactive dashboard, c) post-analysis. The pilot prototype is implemented and tested for real production machinery

The progress of Internet of Things (IoT) leads to appearance of diverse networked environments where specific class of service-oriented applicationssmart spacesare becoming developed and deployed. The deployment is based on certain software platforms, widely accepted as smart spaces middleware. This paper overviews this type of middleware with focus on the requirements that the middleware supports in IoT environments. We analyze and systemize the middleware requirements. The systemized requirements are mapped to the particular existing middleware solutions, leading to identification and comparison of basic approaches in the smart spaces development problem.

Many existing video surveillance solutions are based on sending information between camera and remote server. In this paper, we study video services that are being built in an edge-centric Internet of Things (IoT) environment. Multiple cameras are available. Each video stream is subject to create simple events. Complex events can also be created based on the combination of simple events and context from IoT environment, forming the semantics for video analytics. We introduce microservice-based on step-by-step technology to show how software can be designed and developed for various application areas using multiple cameras. We extend video services that were discussed in our previous works using the developed technology. Proposal smart video services allow to organize smart video monitoring in the following areas: Industrial Internet of Things (IIoT), medicine mHelath services, SmartRoom.

The observed digitalization of society and economy transforms the data from the following three worlds into digital format: a) the information world (accumulated human knowledge); b) the physical world (information from things around people); c) the social world (society and human expertise). Data fusion of these worlds is supported by the Internet of Things (IoT) technology. In this paper, we present an adapted MSc study program for students in Applied Mathematics and Computer Science in Petrozavodsk State University. We show how practical IoT problems become interesting research and development topics for educating the MSc students. The proposed adaptation can be further used for other basic MSc programs.

An equipment monitoring system is responsible for analytical services that assist the personnel in equipment diagnostics, fault prediction, and post-analysis. In this work, we consider a case study of using virtual and augmented reality (VR/AR) for effective analytics delivery to the production equipment personnel. The reference analytical services for our study come from our pilot development with metalcutting industry in Russia. The two service models are proposed: (a) Situation Center and (b) Virtual Assistant. We identify the VR/AR advantages compared with the traditional style of visualizing analytical services on computer screens. Our study is supported with prototype implementation of the VR/AR system for analytics delivery to the personnel.

CNC display (Computer Numerical Control)is an important source of information on industrial production equipment operation (e.g., it shows errors in real-time). In this paper, we consider the direct monitoring problem for a CNC display to automatically detect error codes and analyze their sequences. We present our pilot service implementation, which uses real-life production equipmenta VARIA machine with CNC SINUMERIK 840D sl system.

This work introduces the semantic network construction scheme on top of museum collection and other Information Sources. The problem of museum collections is reduced to weak description of exhibits. The earlier developed ontological model is provides rules the structural presentation of information. Semantic description about exhibits is represented as a semantic network formed according to ontological model. Such a structured and related presentation of information allows you to apply ranking algorithms for its identification and selection. The semantic network is used to build digital services to support the museum. Software agents interact with various data sources to construction a semantic network.

In the age of neural networks and the Internet of Things (IoT), the search for new neural network architectures capable of operating on devices with limited computing power and small memory size is becoming an urgent agenda [...]

Many existing video services are based on trans- ferring the data stream from a video capture device to some powerful data processing center (e.g., cloud-based). In this paper, we consider video services constructed in an edge-centric Internet of Things (IoT) environment. Several edge video capture devices are available. Each data stream is subject to local processing for event detection. Complex events can be also locally detected based on the observed context situation in the IoT environment, forming the semantics for video-analytics. We present a concept model of an edge-centric service-oriented system and its event- driven model for local video data processing. In our demo imple- mentation, we consider a video service of personnel recognition around a production equipment unit, which is used in digital monitoring the smart manufacturing process.

The paper introduces an approach to collecting and mining health-related information on the patient based on sensed data from various medical devices as well as from other digitallyenabled sources. The regularly sensed data are semantically linked thus creating an additional information space-semantic layer. On the latter, a linked knowledge-rich structure-semantic network-is maintained and used to construct mobile services. The use of various medical devices and other data sources makes it possible to remotely monitor patients' vital physiological parameters and other important health-related events. It includes sensing the context of the physical environment, which is then coupled with the health state of the patient. Several patients and interested people can be virtually integrated into a group. Consequently, social methods can be used for enhancing the treatment adherence and for motivating the healthcare goals. As such, the healthcare services would have became more focused on the patients and their needs. Ultimately, the mobile healthcare moves towards the vision of At-Home Laboratory (AHL) that diminishes the necessity to visit the hospital and to directly use its facility.

Performance fluctuation on the semantic information broker side due to critical growth of the request queue length can cause its complete failure or substantial degradation of its services.Since semantic broker is one of the key element of the smart space architecture its performance is a vital issue for the whole environment. We consider two possible sources of the performance degradation.One of them is the growth of the request arrival rate beyond the limits of its capacity. Another source ?nds its origin in the request synchronization which happens due to the presence of the regular patterns in the activities of the smart space participants. Rather often implementation of the relevant approaches at SIB size poses substantial dif?culties. We propose several algorithms that can be implemented on the KPs side. The algorithms form an additional mechanism that can reduce SIB fault rate and avoid performance degradation by regulating the KPs activity and hence balancing the load of SIB. The mechanism presumes additional timeout to the persistent request. The timeout value is selected according to the current system state and its history in the nearest past. The timeout consists of two components. The ?rst one is determined be the active control with the adaptive strategy and the second one is defined by the random backoff algorithm.

Healthcare digitalization requires effective applications of human sensors, when various parameters of the human body are instantly monitored in everyday life due to the Internet of Things (IoT). In particular, machine learning (ML) sensors for the prompt diagnosis of COVID-19 are an important option for IoT application in healthcare and ambient assisted living (AAL). Determining a COVID-19 infected status with various diagnostic tests and imaging results is costly and time-consuming. This study provides a fast, reliable and cost-effective alternative tool for the diagnosis of COVID-19 based on the routine blood values (RBVs) measured at admission. The dataset of the study consists of a total of 5296 patients with the same number of negative and positive COVID-19 test results and 51 routine blood values. In this study, 13 popular classifier machine learning models and the LogNNet neural network model were exanimated. The most successful classifier model in terms of time and accuracy in the detection of the disease was the histogram-based gradient boosting (HGB) (accuracy: 100%, time: 6.39 sec). The HGB classifier identified the 11 most important features (LDL, cholesterol, HDL-C, MCHC, triglyceride, amylase, UA, LDH, CK-MB, ALP and MCH) to detect the disease with 100% accuracy. In addition, the importance of single, double and triple combinations of these features in the diagnosis of the disease was discussed. We propose to use these 11 features and their binary combinations as important biomarkers for ML sensors in the diagnosis of the disease, supporting edge computing on Arduino and cloud IoT service.

In mobile traffic transmission networks, the limited allotted spectrum-involving voice communication and data communication-is faced with the problem of limited bandwidth and consequent traffic congestion. The problem leads to poor communication services like incomplete calls, call drops, slower Internet speed, higher cost of internet service, and other undesirable effects by the service providers. The bandwidth saturation and congestion conundrum still remains critical research area, especially with emerging multitude of edge and intermediate Internet devices. Due to introduction of fog computing such devices become active participants of mobile traffic transmission. In this paper, we consider the Data Interleaving Technique in Mobile Communication (DITMC) and introduce the DITMC-based methodology for reducing the data traffic congestion and bandwidth saturation in GSM networks. The methodology is supported with an Erlang formula model, and our simulation experiments indicate the efficiency. We show that the use of methodology can be extended to other Internet problems like multi-path routing of data streams, active subscription control in smart spaces, and possibility to delegate supplementary processing to edge devices.

Nowadays there is a need to receive and process large video streams in shortest time. This can be realized by interaction of video cameras with additional components of the environment (IoT-devices). Thus, third-party devices will be involved, allowing to simultaneously calculate large amounts of data. Applied methods and calculations will be intelligent due to the increase effectiveness of video information management and the use of other peripheral devices for processing. Intellectualization technology can be used in many applied areas. So, in medicine, you can monitor the patient at home using reflective elements (Parkinson's disease). Vision systems allow you to recognize different information based on video data. In the industry, monitoring tasks arise production equipment directly by the employee (for example, identification of deviations in the work).

Humans loose functionality of many physiological systems and, hence, the social adaptation, after the age of 60. 65 years. The motor, cognitive, and circulation systems are the most essential for personal social adaptation. The human brain adapts to these age-related dysfunctions by means of natural reorganization and modification of its activity, though non-sufficiently. In this paper we review possible “external” interventions based on the Ambient Intelligence (AmI) technology. Interventions aim at compensation for growing social de-adaptation of the elderly that takes place due to cognitive and motor impairments. We propose and discuss a novel concept of AmI-based “supplemen-tary neocortex” (neo-neocortex) or “augmented intelligence” that in a close-to-natural form would partly substitute declining own cognition and mobility. We introduce our design vision of At-Home Laboratory (AHL) assisting the elder person posted on the real state of his/her functionality in everyday life.

A smart space enhances a networked computing environment by enabling information sharing for a multitude of local digital devices and global resources from the Internet. We consider the M3 architecture (multi-device, multi-vendor, multi-domain) for creating smart spaces, which integrates technologies from two innovative concepts: the Semantic Web and the Internet of Things. Our research focus is on analyses of the capabilities of Smart-M3 platform, which provides software implementations for such a central element of an M3 smart space as Semantic Information Broker (SIB). The paper presents a state-of-the-art and contributes our systematized vision on the SIB design and implementation. The analyzed open source SIB implementations include the original Smart-M3 piglet-based SIB, its optimized descendant RedSIB, OSGi SIB for Java devices, pySIB for Python devices, and CuteSIB for Qt devices. We also analyze the design of proprietary or incomplete SIB implementations: RIBS for embedded devices and ADK SIB built upon the OSGi framework with integration in the Eclipse Integrated Development Environment. The theoretical study is augmented with experimental evaluation of available SIB implementations.

Smart space supports interaction of multiple participants though information sharing. This kind of interaction is indirect and networked. For detection of changes, which come in parallel from other participants, a participant applies the subscription operation. It allows receiving notifications passively on a regular basis. Absolutely dependable delivery of notifications cannot be guaranteed in modern computing environments. In particular, wireless networks are subject of frequent faults, making the subscription operation less reliable for mobile clients in smart spaces. In this paper, we study the problem of active control by a client in receiving subscription notifications. We introduce a simple mathematical model for control the check interval for subscription notifications, which ensures client adaptation to the notification loss rate.

The SmartRoom system provides a set of services for assisting collaborative activity when many participants are physically present in the same room. The SmartRoom service intelligence is due to the use of diverse sources of information on ongoing activity. One promising source is information on user presence and network activity of the participants. In this paper, we consider the problem of utilizing such an information source for smart spaces based on the Smart-M3 platform. We employ the Innorange technology, which is one of the many examples for passive radio detection in WLAN. We define scenarios with this information source for use in SmartRoom services. We propose an integration solution oriented to Smart- M3 applications. Our software prototype is integrated into the SmartRoom system. Experimental study confirms the feasibility and reasonable performance of the proposed solution.

The Smart-M3 platform allows constructing software applications consisting of agents that interact by sharing information in a smart space. An important problem is achieving the dependability of applications in case of failures, a common situation for existing networked environments. The most sensitive operation is subscription-a persistent networked query. In this paper, we consider a generic software infrastructure for a Smart-M3 application and propose infrastructure-level solutions to support the application fault tolerance. Our first solution augments the infrastructure with a content service. It improves safety (preservation) and integrity of volumetric data due to the delegation of storage functions to a separate element of the application infrastructure. The service employs the wellknown matured web technology. The second solution consists of mechanisms for the fault tolerance support to detect failures of subscription and to recover effectively. For case study, we use an existing Smart-M3 application-the SmartRoom system. Our evaluation confirms that our solutions lead to better dependability without essential degradation of the performance.

SmartRoom is a system supporting collaboration activities localized in a room: a set of digital services is available for organizers and participants. The Smart-M3 platform is used to set up a networked knowledge sharing environment on top of which the SmartRoom service set is deployed. In this paper we introduce the methodology and design of the SmartRoom system. We consider realization of digital environment in a room, service set structure and corresponding Smart-M3 based computing infrastructure, ontological models and their use for service construction and delivery.

Smart Room provides a service set to automate such research and educational activity as conferences, lectures, and meetings. For human participants the services are accessible via mobile personal end-user devices. A mobile device hosts a smart room client, which registers the participant in the smart room space, shares her/his personal data and context, and accesses available services in the room. This paper reports recent progress of the smart room client development. Our key objectives are (1) development unification for various mobile platforms, where Windows Phone and Symbian are our reference cases; (2) support in the user interface of personalized access to the important services from a large set. Our initial study shows that objective (1) is an architecture design issue and our multi-platform development is based on tradeoffs between native mechanisms and portable solutions. We achieve objective (2) using HTML5 features, supported by many modern mobile platforms.

Activity tracking in smart space applications allows extending the applications with new features, such as automated generation of activity reports. In this paper we present a generic system - Event Recorder - that potentially can be used with any smart space application. We consider a particular case for Event Recorder: tracking events and producing summary reports of the activity held in Smart Room. The contribution of this study includes use case scenario and ontological representation model of Event Recorder, integration scheme of Event Recorder into the Smart Room system, and evaluation of appropriate visual formats for summary reports.

Petrozavodsk State University (PetrSU) has started development of smart rooms aiming at making the PetrSU educational and research processes of a higher automation and quality level. An important component of such rooms is sensors, which allow monitoring the current physical environment state. This paper reports our work-in-progress results on sensor use in PetrSU smart rooms. First, we selected classes of sensors for use in PetrSU smart rooms. Second, we introduce an ontological model for representing sensor-related data in smart spaces. Third, we propose an architectural Smart-M3-based solution for connecting sensors to the smart room space. We also discuss basic scenarios with sensing data that we plan to implement in the smart room.

The Smart-M3 platform allows prototyping and experimenting with various smart space applications. The performance problem of such applications and the level of automation in the development process is, however, in its early phase. In this paper, we evaluate the knowledge pattern mechanism that the SmartSlog SDK provides for Smart-M3 applications. On one hand, the mechanism allows writing compact code to implement advanced search queries to the smart space. On the other hand, high-level mechanisms are for the cost of performance; it can essentially degrade when the size of search problem grows. Therefore, a tradeoff has to be determined. Our initial experimental study considers a worst case model of the knowledge pattern mechanism and we provide empirical performance upper bounds.