0% found this document useful (0 votes)
8 views8 pages

Smart Health

The document discusses a complex event-driven data analysis approach in fog architecture for smart healthcare, focusing on managing and processing large volumes of IoT-generated data. It highlights the use of Complex Event Processing (CEP) technology to extract critical insights in real-time, thereby reducing unnecessary data transfer to cloud servers and improving efficiency in healthcare applications. The research emphasizes the design of a computationally efficient analytic engine that enhances patient health monitoring through online ECG analysis.

Uploaded by

Swapnaja Hiray
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
8 views8 pages

Smart Health

The document discusses a complex event-driven data analysis approach in fog architecture for smart healthcare, focusing on managing and processing large volumes of IoT-generated data. It highlights the use of Complex Event Processing (CEP) technology to extract critical insights in real-time, thereby reducing unnecessary data transfer to cloud servers and improving efficiency in healthcare applications. The research emphasizes the design of a computationally efficient analytic engine that enhances patient health monitoring through online ECG analysis.

Uploaded by

Swapnaja Hiray
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 8
2022 EE 3d Glbal Conference for Advancement n Technology (GCAT) | 978-1-6654-6855-8/22/531.0 2022 IEE | DO: 10.1109/GCATSS367:2022.9971877 2022 IEEE 3rd Global Conference for Advancement in Technology (GCAT) Bangalore, India. Oct 7-3, 2022 Complex Event driven data analysis approach in fog architecture based Smart healthcare Swapnaja Rajesh Hiray PAD Scholar CSE GITAM Visakhapattanam and PICT Pune. India [email protected] Abstract: Managing and processing the huge amount of heterogeneous data generated from IoT Is the most difficult task. It called info question the computational storage capacity and efficiency of today's infrastructure resources. Moreover, it becomes dffieult to provide the user with sigificant information to make a decision due to the existence of the massive data. Complex Event Processing (CEP) technology has developed as an effective method for extracting critical insights from enormous streams of data in real time, Here an attempt is made to design a system that can generate complex events from analysis of large sensor data and send only critical events (o the cloud server. We have included data analytic engine on the e, dge device which will (ake advantage of CEP technology and considerable reduction in data volume is carried out as We are pushing only Complex events, Our research mainly focuses on the design of this computationally efficient analytic engine for the data reduction in Smart healthcare applications, It avoids sending unnecessary event data on the cloud; subsequently, reduce the data transfer time and memory space. We have developed hierarchical event riodel on the edges in smart health care application and used the fuzzy based approach for analytical processor design on the edges. Application mainly focuses on the online ECG monitoring forthe critical patients. Keywords—CEP, Fog Computing, Smart Healtheare, Fuszy Logie, Inference Engine, CNN. 1. Ivtropuctiox Making the use of ToT in the healthcare industry takes it to the next-generation smart digital environments and remote monitoring. Despite the potential advantages of a remote monitoring system using wearable's, there are significant challenges ahead. Sensors in the Internet of Things ecosystem create a vast volume of stream data, Analyzing such huge data generated by the sensor is the main challenge that needs to be addressed, that prevents from obtaining accurate results, Generally, the data getting from the sensor is uncertain, so analyzing such uncertain data streams (1 in real-time is quite difficult, Due to the lack of appropriate solutions in the eurrent state of the art, there is a need for a system that evaluates only important content from the continuous stream of data and sends only significant data over the cloud to reduce computational complexity. It is critical to have a system that can analyze and deliver a comprehensive analysis of data with greater accuracy for real-time decision making on streaming data. Different techniques are available [1] [2] for extracting valuable {97¥-1-6654.6855-8122'31.00 62022 1 Dr, Bhremaramiba Ravi Professor CSE GITAM Visakhapatnam, Andhra Pradesh India ravi gitam.ed {information from continuous data for post-event analysis that recognize the anomalies while they are occurring. Such detection leads to having @ considerable impact on the event outcome, ‘Complex Event Processing (CEP) is an emerging study subject that has the ability to assess data in order to discover ‘unobservable behaviours. It reflects more relevant occurrences from several sensors than the plain event, which aids in giving, real-time information, For example, an arrhythmia event may be recognized if the respiration rate increases while the heart rate exceeds the threshold limit within a given time frame. However, present complicated event-processing systems take a long time because they do not account for sensor data similarity and redundancy. This system attempts to transfer all basic data to the cloud, which complicates the system and ‘causes it to take too long to upload the data. There are multiple techniques in the literature that give prediction answers in various areas, such as [3], which detects Complex Event Processing (CEP) utilizing rule-based machine learning methodologies. Furthermore, efforts have previously been made to create CEP using Esper, as Boubeta et al. [4] mix technologies such as Mule ESB and ESPER engine utilizing the Xively IoT platform. However, owing to the centralized framework, it does not provide the dynamic communication protocol necessary for event-driven architectures, and it does not benefit from multi-cloud environments The primary goal of this study is to evaluate large amounts of streaming dala in order to acquire real-time insights that have the ability to accurately forecast patient health and send ‘warnings to doctors. Creating complicated events on a mobile device and delivering that data to the cloud rather shan streams ‘of big data from several sensors may assist to save data transfer time and storage space, The suggested CEP computing strategy for data analysis in healthcare applications addresses ‘outstanding issues in this regard, These occurrences aid in the prediction of patients’ health and give assistance for prompt responses. A machine leaming method is used to enhance prediction of a patient's health (Cloud based CNN), I. LITERATUREREVIEW A. Smart healthcare Architecture (lo) Smart Health isthe use of information and communication technology to increase availability and quality of health eare With a growing elderly population and rising healthcare ‘AutrarandIeonsed use Inte to GITAM Universty,Oounloades on Drcembe 28,2022 x 1123-54 UTC tom IEEE Xplore, Restrctons spy. expenditures, this topic has drawn the attention of both scholars and healtheare professionals. Current health-care systems are overwhelmed and, as a result, cannot meet the growing population's demand, With the widespread availablity ‘of mobile phones and health trackers [3][S] capable of Capturing info on people's health in real time (ECGs, temperature, body oxygen saturation, and other biosensors), as well as by monitoring everyday ‘activity and identifying ‘anomalous motions using inertial sensors, itis now feasible to evaluate this data in the cloud to make more informed healthcare decisions. As a result, total expenses and the stress ‘on healtacare institutions are reduced. The illustration depicts ‘a smatt city healthcare system powered by IoT, Refer Table I. TABLEL, MAIN CHALLENGES AND ADVANTAGES OF A MODERN HEALTHCARE DELIVERY SYSTEM. Chalenes Benepe ‘Sealy ond roperaiy pred ccesty sent at we equipment Cost sang eit and Priaey fortes use B. Fog computing in lo Healthcare system Fog computing [14] [17] refers to the processing of measured data by sensors on sensor devices or devices adjacent to sensors rather than on cloud servers or faraway healtheare centers. So, in essence, an upgraded fog computing technique is @ low-latency diagnosis or alarm mechanism. In contrast to cloud frameworks, which are carefully focused, fog computing allows an organization to use cloud administrations at the organization level, but it also allows the organization to use local computational resources at the organization level. The fog nodes also assist (0 save energy by limiting long-distance transmissions and selecting neighbours based on a time restriction calculation, ©. Complex Event Generation Fog architectures for real-time [oT applications were studied by Giovanny Mondragon-Ruiz et al. [67]. Researchers found that these designs increase resource allocation while ‘minimizing latency with CEP-engines' suitability for low-cost computing, Fog computing decreases latency by up w 35 percent compared to cloud computing, and low-cost devices like the Raspberry Pi have shown to be capable of delivering realtime loT applications with the quality of service. For CEP related literature Survey please refer Table 2. Like sccurity surveillance and healthcare systems in real-time environment, cloud computing presents a number of unique issues. Sccurity and privacy are critical concems when it comes to healtheare oT. The main problem observed here is that data processing is proportionate to the Data generated and transferred throughout the network. So if we want to reduce this, we should think about reducing the data resources only, So subsequent routing of data over the web will be avoided. So here, we found the research gap, and then we have developed the innovative rule- based complex event-generated engine on edge devices.We did the comparative analysis for data reduction with the help of various Machine learning algorithms. We found that time and space complexity is the main problem with these resource ‘constrained edge devices. And then come up with a Complex ‘event generation module. IL, RESEARCH PROBLEM AND METHODOLO A. Mathematical model for identifying the GAP Basically when we are working with the Smart city applications and talking about Fog architecture and we have following layers present 1. End devices such as Smart phones, RFID tags , Smart watches , In our case we have I, body sensors . 2, Hdge devices such as routers, gateways, switches, and access points. These are connecied to end devices with any of LAN like protocols 3. Cloud layer where major servers and processing is going on . Generally when we are transferring data to edge devices from end users we are using LAN like protocol where communication delay. is considered negligible. Also while modelling the network for getting performance evaluation parameters Unprocessed data in smart applications is forwarded to cloud servers via intemet for processing so communication delay from edge devices to cloud servers needs to be considered. There are certain assumptions when we are discussing Fog Computing architectures for IoT smart city applications: 1. The terminal nodes (TNs) in the network, such as mobile phones, Body Sensors are aware of and able to transmit their absolute geo-spatial location, Which is further used in the ‘event modelling. 2, The fog computing tier where edge devices such as Routers data collector nodes are present comprises of ‘intelligent’ devices which are capable of computing, processing, and storing data in addition to routing and forwarding the data-packets to the upper tier which is Cloud layer, ‘AutrarandIeonsed use Inte to GITAM Universty,Oounloades on Drcembe 28,2022 x 1123-54 UTC tom IEEE Xplore, Restrctons spy. 3. The networking devices in the fog computing layer are able to share the network, computational, and storage lod among themselves as per requirement 4. All the communication network layer delays handled by standard protocols, {In our Smart healthcare application the tier closest to the ground builds a network of several body sensor-cquipped, Internet-connected end-devices, often termed as [oT The data ‘transmitted by these sensors which are called as the Terminal nodes are received by the edge gateways present al the edges of the fog tier. Unlike the traditional cloud architecture, in fog computing not every data packet is redirected to core cloud computing module for processing, Instead, all real-time analysis and latency-sensitive applications are run on tae fog layer itself. The fog computing devices in this layer have limited semi-permanent storage that allows ‘them to store the received data temporarily for analysis and then send the source devices the needful feedbacks. When we try to model the Smart healtheare application there are two parameters contributing mainly on the end to end delay to get the results. In smart city healtheare applications, traditionally data processing delay or computational overhead mainly depend on following factors:1.Communication delay for data transfer from devices at edge to cloud server. 2.Machine learning computational overhead at cloud server. Following points like the Network protocols , Speed , type of data we are transferring that we are considering standard and constant. That means end to end delay due to these parameters will be there but analytic engine we are designing mainly to reduce the events data transfer to and from cloud. We can graphically model the problem as follows (Ref. Fig.1.) a a a a a os s a 1, Diagrammatialy Representation of Fo architecture We can model graphically above network as follows (Ref. Fig2) Fig. 2. Graphical representation for Fo network 3 With the above said architecture the performance metrics of the Smart healthcare applications can be modelled or benchmarked on the following criteria I. Service latency Service latency for a request sent by an application instance running within an JoT device is basically its response time, and is computed as the sum of the tansmission latency and the processing latency for the request. ‘We assume that communication among multiple edge devices and. between the cloud server is on high bandwidth leading to negligible delay. We assume that communication among multiple Fls at tier 2, and that among different cloud data-centres at tier 3 take place over high-bandwidth channel — leading to negligible delay. Moreover, the communication delay among the TNs is taken (© be insignificant . Let 8if and Sle be the delays in transmission of a data packet from a loT biosensor to the corresponding FI, and from an FI to the cloud data-center, respectively. Thus, the mean tansmission latency, 8fog , for the data packets of N application instances running within is iziven by Blog =[SI0Bi | Bfebi | Ble Bir | StF Bir | SUN BI bie Bi Here, bi_is the cumulative packets send by Ni application instances to Fog router and from fog to cloud respectively. Xr indicates the total number of data packets that are sent ‘sa response to X request data packets, Processing latency for fan application instance request is computed in terms of the number of requests that are processed at the server end prior to its service, Here we can say that in transmission latency no of data packets are directly participating. With only. cloud architecture without fog , with efficient algorithms with AI and ML advancement we can have processing time latency related to service can be reduced 2. Energy Optimization : The energy expended due to transmission of unit byte of data from the bottom tier to the ‘middle tier, and from the middle tier to the cloud tier . But here ‘we are focusing only on latency so detail formulation related to Energy Optimization is beyond seope ofthe paper. ‘Mathematical Model: we want (o restrict our problem for data processing. oa the edge devices and in above scenario can be formulated as follows: Fig No 2. graph consists of m edge devices , like Gateways ,Access points , LAN routers ete V=Qe Vo yk bem} (1) is a vertex set, Where 2i is a set of edge devices and mis the number of edge devives, BR(e 21 2QyneZhehnezm-l,zm} (2) Is edge set ¢ {elziz))} is the communication link between edge nodes zi andzj ‘zi, on the edge is the communication delay between edge nodes Zi and zj,and the, size of its value is randomly generated latency part of the data processing, AS number of ‘computations increases the computational delay will be increased al the device, DEdge i= (1 Vzi*(aix2 0, ‘AutroizedIeonsed use Ivied te ITAM Univorsiy.Downlbaded on December 28,2022 a 1128-54 UTC tom IEEE Xplore. Reston app. where Vzi is the device computing capability of i, xt is device data be processed by é, and ai is a predetermined real ‘number between 0 and 1, So, the computation delay of m edge nodes can be expressed as DEdge = minE mi=1 (1 Vzi) * (sixi2) Where xi function of the amount of data generate at edge layer. delay in cloud computing layer includes processing delay plus delay in transmitting data from the fog layer to the Cloud server layer, For the cloud server j ifthe amount of data it needs to deal with is yj and the computing capability is Vj its computational delay can be expressed as : Dj = yj / vj G= 1, 2....m). so form cloud servers with total data to process is Y we ean have ‘The Communication Delay at cloud layer can be modeled as follows : If the data loss rate is not considered, the delay of | the WAN transmission path from the edge device i to the cloud server jis dij, and the traffic rate is 2ij the delay function can be defined as, Dij = dij j (i= 12, san) and j= 1.2 3jceeeeeat) ‘Therefore the delay in transmitting data from edge devices to cloud servers can be expressed as jl dij aij ‘The system delay in Smart healthcare application with Fog architecture mainly consists of Delay System =D Edge + DCloud + D edge-cloud Where D Edge Computational Delay at Edges , D Cloud Computational Delay at cloud and D edge-cloud Communication Delay From edge to cloud. Dedge-cloud = So our Objective function parameters mainly focus on following : 1, Data amount transferred from Terminal Devices to Edge 2, Data amount transferred from edge to cloud 3, Edge layer data analytics strategies and algorithms which have better Computational performance With this as the GAP analysis and Objective function our Research Objectives are as Follows: ‘© To reduce transmission data packets with introducing suitable methodology on the edge devices. © To do the performance analysis of new suggested algorithm As it will be designed at the fog layer which hhas limited resources © To devise an efficient and novel architectural framework and fine-tuned stacked data analysis approaches for smart healtheare applications which further be integrated with any smart city application B. Methdologies We have suggested the use of Complex Event Processing algorithms on the Edge devices to reduce the amount of data ‘uansferred fiom Edge devices o cloud server Proposed System: We have proposed the following components at different layers along with standard fog architecture in the Smart healtheare application, To discuss the feasibility and analyze the different design paramecers we have used the prototyping approach, Use case we have selected is predicting the heart disease severity for critical heart patient. ‘This architectural design can be extended to various smart city applications by customizing the component design Fig. 3. Proposed archiecturl Design for smart healthcare Design issues and algorithms suggested at the Terminal devices : The main concem while taking the input in any ‘Smart city applications is its heterogeneous environment . For ‘example here we have taken ECG signals , SpO2 and Body temperature thermometer. We have suggested a standard Event ‘model for any type of the sensor . While running the input ‘model on terminal devices where sensors are connected. input port will be getting value every sensor is having standard Device ID , and location so event Object will have following format: € = (UD, DevicelD, DeviceType, Location, StartTime, EndTime, KeyValueList) where ID : Event No , Device ID : 1,2,3,......i (For i different sensors ) , Location : Geographical location , Stat time : time stamp for stat of reading input port, End time : time stamp for start of reading input port , Key values ist: Input port data (It depend on the Sensor ) This standard Object format for storing ‘event format helped us to reduce the heterogeneity. Further this ‘Object oriented event data is used for further analysis Event Collection system : This Block is proposed at the data collector node . Now it depends oa your basic network architecture and WSN topologies you are using, some times it is present at your access point or your terminal device like mobile phone or fit band or a separate data collector node or ‘your router acting as data collector , Ths is like preprocessing ‘you are doing with input data. It is using event filtering and for ‘that purpose standard techniques such as thresholding or ‘windowing can be used . Here we are again different type of ‘vents . 1. Simple atomic events or 2. Composite events . Now the Out put event Object will have following format: + BL (1,16 Value) : Distorted ECG + E2(24kt,Value) : SPO2 below 95 + E3(3,Lvalue) temperature above 96.6 In our use case it ean be: { E2( 2, Tew bed 1,10,00am, 85) 3(3, leu bed 1,10.00am,34) 4 ‘AutroizedIeonsed use Ivied te ITAM Univorsiy.Downlbaded on December 28,2022 a 1128-54 UTC tom IEEE Xplore. Reston app. EI(1 Jeu bed 1, 10.00am, 389404297) } Conceptual diagram for our event model ean be Fig 4 Proposed Event Mode In the proposed event model We have developed the hierarchical event model which is in line with our objectives and pushing only significant events which are service and application oriented. The business rules for this are proposed by domain experts . In our case we have contacted the practicing cardiologist to get the correct rules for significant events Fig 5. Gives the Simple event generation algorithm. Now this algorithm is producing input to our Fuzzy based Complex event generator . Which are further pushed on the cloud for further data analysis, ie 5 ‘mpl Event Processing Algritim Complex Event Processing systems CEP: - It represents more meaningful events coming from different sensors compared to the simple event, which helps in providing real- time insights. As we know that any smartcity application is pervasive application hence it has Context-awareness as onc of the important properties. Complex event processor which is @ ‘very important component we have suggested for further data reductions. The system uses complex event processing (CEP) ‘to detect the abnormality in patient's health condition. Complex events (CE) are generated by using the fuzzy based approach. This is third Novelty we added in our component design at the edge device. It generates alerts in realtime as a complex event and pushed to cloud with ECG values for further trained model Fuzzy based approach for design of CEP: As we can see in smart-healtheare problem statement — ECG and Spo2, Temperature are analysed and thresholding technique is 5 used for filtering and creating the critical events , data transfer in bytes is considerably reduced . This we can do offline also ‘with low intemet connectivity. But still the set of values (Sensor reading) is sharp boundary values. With fuzzy rule base here we are allowing smooth boundaries and then applying rule-based inference engine with simple and ior ‘operation to generate new complex events and which will carry ‘only necessary information used for cloud based analysis. This ‘complex events are pushed further. Fuzzy set concept is applied on atomic events which we are getting as the output from our first algorithm and as per rule complex event will be sgenetated. In this fuzzy representation this limitation of sharp boundary set further analysed by allowing membership in a set to be matter of degree . the degree of membership in a set is expressed by a number between 0 and 1 ; 0 means entirely not in the set , I means completely in the setin between means partially in set. From Fig 6 Its clear that In put to our block is Critical events . Fuzzification functions we are using are triangular and trapezoidal. Following is CEP Algorithm Input data set : SE(4,Ltvalue)! Where d=(1,2.3.45), |= location(bedno),t=time stamp, (Atomic events after filtration) Output dataset: BELEE2,, 1 then( call fuzzification function for ECG) 2 then (call fuzzifieation function for SPo2) 3 then ( call fuzzification function for BT) 4, Forall events with same I and t Search for patterns in Ruleset 5. Generate complex event when Rule Matches 6. Push Events for next analysis to Cloud Fig. 6. Simple Complex Event Processor Vv |As stated above objectives we have developed the prototype architecture with heart discase prediction application. Usecase for our prototype consists of three Biosensors ECG, SPo2 and Temperature. As it is very difficult to directly use sensors with prototype, we have stored the sensor values in a text file for each type of sensors as tuples of values and used those text files as the input files. We have used AWS cloud instances for the patient node, Edge node and cloud node. RESULTS AND EXPERIMENTATION ‘AutroizedIeonsed use Ivied te ITAM Univorsiy.Downlbaded on December 28,2022 a 1128-54 UTC tom IEEE Xplore. Reston app. Sa Spot Fig 7. Expeimental setup. 1, Patient data transfer on the cloud with reduced data volume with data analyties model 2. Performance efficient data analytics Complex Event Processor on the edge device 3. Suggesting the suitable data analytics framework for ‘Smart Healthcare service with fog architecture, Model 1: - Patient Data transfer on the cloud with data analyties model + Input: - Sensor Generated Data Temperature sensors, oxygen saturation sensors, and clectrocardiogram sensors are among the sensors that the proposed system uses in real time. The sensor data is the primary input to the system thet is collected using an Arduino UNO microcontroller. An Arduino-like microcontroller receives data from a patients sensor network on a constant basis, Sensing data from the heart, this component delivers it to the gateway devices attached to the system, + Output: = Simple event generation and pushing that on cloud, and Complex events Process: - Here we will be collect the patient's sensor data as a system input. The data files are stored on the cloud respectively. These Raw events will be categorized by our simple event generator analytic engine and high priority events are forwarded. With these high priority events as per rule database and generated context are converted into complex event and necessary data is pushed to cloud server. Fig. File size VS time taken to upload gph Analysis: We ean see time required to upload, the same size data with same no of patients, without CEP is more and with CEP itis dependent on the No. of complex events generated Model 2 : Performance of Complex Event Processor on the edge devices. Input : Events generated from our frst Simple event model EI (1, 1,4, Value) ; Distorted BCG 2 2,)4,Value) : SPO2 below 95 E3(3,L4value) temperature above 96.6 Output: Complex Events CE (Event id,Patient id location , Time stamp , SpO2 values, action 1 ) CF2¢ Fvent_ idsPatient id location , Time stamp , SpO2 values, action 2) CEN ( Event id.Patient id location , Time stamp , SpO2 values, Temp values ,actionN) Process : Here we have developed Fuzzy based nile engine to generate the Complex events . and depending upon complex ‘event generated actions will be taken. As shown in the above ‘output action 1 can be alert generated to Drs. , action 2 ean be heart related event which will be treated as the input to out prototype service where our CNN is running which will be used to detect heart attack, And so on n actions depending on various services. The rules for the Complex event gencration for our protype we have taken are as follows : RULE 1: IF ecg IS less_distorted AND spo2 IS hypoxia? THEN result IS CEI; RULE 2 : IF ecg IS very_distorted AND spo2 IS hypoxia2 ‘THEN result IS CE2; RULE 3 : IF ceg IS very_distorted AND spo? IS hypoxia’ ‘THEN result IS CE3; RULE 4 : IF ecg IS very_distorted OR spo2 IS hypoxia’ THEN result IS CE4; We have used simple and / or functions fiom the CEP ‘operators for our rule base ig. 9, CEP performance Analyse Graphs ‘The processing time Te inereases as the number of complex. ‘event definitions N increases for all of the two kinds of ‘complex events. Elapsed time is the amount of time that passes from the start of an event (o is finish, Here We have M which is Raw Events and last complex event we received. ‘Model 3: Machine learning (CNN) on MIT-BIH dataset ‘+ In case, IF spo2 < 90 AND spo2 >30 THEN © Low Breath Event is generated, + IE temp > 37 Celsius THEN 0 Event is generated ‘Temp 6 ‘AutrarandIeonsed use Inte to GITAM Universty,Oounloades on Drcembe 28,2022 x 1123-54 UTC tom IEEE Xplore, Restrctons spy. © Heart Rate < 50 OR heart Rate >125. abnormality event is generated Heart Rate ‘The ECG signal getting from the sensor containing noise so preprocessing can be done to remove noise and get the final ECG value. Parameter normal ranges are mentioned below. + RR interval: 0.6-1.2 seconds for normal patient) (min-max range PR interval: 121 200 milliseconds. + PR segment: 50-120 milliseconds + QRS complex: 80-100 milliseconds, + ST segment: 80-120 milliseconds. ‘© T wave: 160 milliseconds. (ean ignore if not possible) ‘+ Pwave: 80 milliseconds. (can ignore if not possible) Else Abnormal heart rate If all the sensor values are high (referring rule sed), then complex event(CE2) is generated, these events are seat to the cloud and On cloud a machine learning techniques CNN is applied to predict heat failure and critical alerts are sent to the doctor or family member's. The analytics reports are being generated on cloud, We have sent all the data stream to cloud, for those patients for which the CE2 is generated. The data is compressed and stored on cloud. Compressor algorithm we ‘used standard whieh is beyond scope of problem statement. The overall accuracy obtained for the Classification is 98.16% using CNN. Pee eee meres Fig. 10. Graphical reretenttion ofthe Patient's Data files transfer On the ‘loud server (AWS) V. CONCLUSION In this paper, we proposed a novel computing approach for data Analysis using Complex Event Processing (CEP) & machine learning in the IoT environment. Most of the available Complex event-processing methods take a long time because they do not consider similarity and redundant operations individually concerning basic events. The given system ean be used for remote patient monitoring over a cloud environment, We conducted an empirical study on the memory data to investigate the applicability pastem generating algorithm with business rules. The time required for data transfer is greatly 7 reduced as only complex events are pushed over the cloud instead of the whole sensor data, The support vector machine algorithm used is providing the health prediction for that particular patient and alert the doctor for further treatment, Esper is used for the generation of the CEP engine on the ‘mobile device. The system is able to analyze data & generate a real-time alert RFERENCES. 1] Nist Mebdiye, Julian Krumsich, David Enke, Dirk Werth, Peter Loos, Determination of Rate Pater in Complex Fvent Processing Using Machine Leaming Technighes”, Conference Organizes by Missort University of Seence and Tecnology 2015-San Jose, CA. [2] pe/mteiiescounelcon/aytem tiple snug datasteamsinioe [07491 pile iype-node&id-38628force 5] ijt Mehdyeva, lian Krumscha, David Enksbe, Disk Werta, eter Looss,"Detmiaation of Rule Patems in Complck Event Processing Using’ Machine Learning Teshnigues", Complex Adaptive Systems Pablicaion Cikan H. Dal, Editor in Chie Conference Organized Missouri University of Sctence and Tecboslogy 2015-San Jose, CA {4] Juan Rouhee-Puig Guadalupe Orisimaculada Me Bulo,"Approaching fhe latanet of Things though Integrating SOA and Conpiex Event Processing’, March 201d naps eserehgae nepublieaton 261877908 5] Yao W, Chao-tisen Chu and Li Z 2011 Leveraging complex event processing for sma hospital using RFIDJ. Network “Computer, Appl) 798-810, 9] R. Pathak and V. Vaidehi, “Complex event. procesing_bated remotehelth monitoring system,” snPree. Sd Int, Cont Eco freny Comput.Commun. Sy. ((CECCS), Dos 2015, pp 6-65, 1] A Manas, 1. Light, and U. Yada, “Healthcare event aggregationa UWBAL), 2 knowledge sharing platform for anomaly dleconand prediction” inProc. 17th Int Covnf, Etlealia Netw, Appl Services {HfeatthCom, Boston, MA, USA, Oct, 2015, pp, 688-653, 8] Alef Mabatar, Imac! Bouassda Rodtiguer, Khall Cae, Leila Abi, ton Bema Freisleben “CEDAHFP: Complex Event Procesingfor Hear Frlue Prediction” EEE TRANSACTIONS ON NANOBIOSCIENCE, VOL-16,NO. 8, DECEMBER 2017 9] V. Vaidei, R Bhargavi, K. Ganapethy, and CS. Hemalatha"Wiltisensor ‘based inbome ‘healt monitoring “using complex eveniprocenting,” inact. Cont Recent Trend In, Technol, Ape 2012.9 570-575 Dhillon, S. Mejumdar, M. Stake, and A. Ebtarki, “MCEP:A tmobile" device based complex event processing system for remote bealt-cice, irae IEEE Tot. Cont lvernet Things HCIOT), Jl Aug. LINA LANI2, RUISHENG SIIB, BAI WANG2, LEI ZHANG, AND [NING JIANGS “A Univeral Complex Event Processing Mechanist Based on Edge Computing for Inert of Thiags Real-Time Monitoring” Dipl Object lenier 10, 109/ACCESS 2013,2930313, IEEE Access-Augest 201 ‘lagen Dunkel, Ralf Bras, Ssbastian Stipkovie. “Event-Based Smartphone Sensor Processing for Ambient Assisted Living” 2013 [EFE: Floventh international Symposium on Autonomous Decentalized Systems (ISADS) .Bade(2010)Esper Android EveatweunProcessiagonAndeod [Onlin] Avalblehtps ‘sow informal unichamburg.elpojet exper anol 1. Amide (2009) Joining Oracle Complex Event Processing and JOMEt0" Reast to. Location “and Portioning “Evens, [Online Avaldble Link: -upfwww.oracle con/techactwodvarclesamade! 6p-090895 ul J, Dunkel, “On Complex Event Procesing for Sensor Networks”, in Proceedings of the Intemational Symposia on. Actonomous Decentralized Systems (ISADS), pp. 249-284, (FEE (2000), Mobored, A. Mista, M. Ebling apd W. Jerome. HARMONI: Context aware Fileing of Sensor Date for Continuous Remote Malth esourceslundrst 10) uy) 12) 3 144 ‘AutrarandIeonsed use Inte to GITAM Universty,Oounloades on Drcembe 28,2022 x 1123-54 UTC tom IEEE Xplore, Restrctons spy. v3) Ls) bs) 20) bay pa) Monitoring. Tn Proceediags of Pervasive Computing and Commaneation (PerCom), pp. 248, TEBE (2008), 978-0-7695.311327. ‘A to, T. Oma, ¥_ Kakuda, . nove, Sefety Support System on School ‘Routt Based on Grouping of Chilea ia Moble Adios Networks, ix Proceedings of the nicnationalSynposium on Autonomeus Desenzalzed System (ISADS}, p.885:538, IEEE Q2011), ‘A. Moutham, L. Peyton, B. Eze and A. El SaddkEvet DrivenDats Integration for Peronal Health Monitoring, Joural of EmeigigTecholopies in Web tlligene, voll, 202, pp. 144 148, Novembe2009, S. Reddy, V. Samanta, J. Burke, D. Vsti, M. Hansen, M. Svartava;MobiSense = Mobile Network Servos for Coordinated Parcipstry Sensing, in Proceedings ofthe Ilerational Symposium ot ‘Autonomous Desentalied Systems (SADS), pp. 1-6, IEEE 2009), 1. Dubos, J. Yang, N. Constnt, A.M. Amin, Q. Yang, and K. Makodiya, "Fog Date: Enhancing Telehealth Big Data Through Fog Computing.” in Proceedngs of the ASE BigDaua & Sociallformatcs 2015, New "York, NY, USA, Oct 2015, pp. 1-6, dol iW. 114si281 ess 2418889, Hossain MS, Muhammad G (2014) Cloud-sed collaborative media service framework for heabsare. Int 1 Distributed. Sensor Netw 0G 888712 YYoussfpour, A Isigaki, Gi Jue, JP, Fog Computing: Towards [Minimizing Delay in the Ineret of Things, In Proceeding of the 2017 TELE st lnvernational Conference on Edge Computing, Hono, #1 USA, 25-30 Jue 2017p. 17-2450, ® 3) ea) 25) (25) (7) a Aff Malta, Ismael Bowassida Rodrigue, Khalil Cari, Lela Abid, toa Bernd Freslchen “CEPSHEP: Complee Event Processing for Hear Failure Priston” IEEE TRANSACTIONS ON NANOBIOSCIENCE, VOL. 16, NO. 8, DECEMBER 2017 ‘Alieza Manashiy.J. Light "Cloud Patforms for 1 Healthcare Context ‘Awareness and Knowledge Sharing’, nasty 2017 DOL 10.10079783- 519.50788-8-12 In book: Beyond the Inlet of Things (pp. 303-322) [Azim A. Anzanpour, AM. Rahman. Peitiala, M. Levorto, P. Lileberg, No Dut, HCH: hierarchical og-asisted computing acitecie for healthcare ToT, ACM Trans. Embed. Comput. Syst (ECS) 1659) 2017) 174 Kau ad Jsuja Monitoring pulse rate, body temperatueTempecature sensory and heat ae seasoe Respoeny FiRemote allt ‘monitoring by using Bluemix cloud, Magata Espinoza tal, Detecting and alerting professionals about pesons filing to the ground; even-based mmontoring to. report Uuehyeardia and bradycardia, Heart rate sensor and tiple axis fcclerometer LCP2148 ARMT microconrller Encryption" acheme far wireless sensors commramication. Mobile application and web page foreasy acess end push noiestions of abnormal evens Orbs, S.Oniga, Automated system for evaluating heath status, design tnd eehuology in elecvoule packaging (SUTME), ia: IEEE 19th Intentional Sym postu fr, 2013, pp. 219-222 ‘AutroizedIeonsed use Ivied te ITAM Univorsiy.Downlbaded on December 28,2022 a 1128-54 UTC tom IEEE Xplore. Reston app.

You might also like