logseq/pages/hls__Najdataei - Efficient Data Streaming Analytic Designs for Para.md

file:: [Najdataei - Efficient Data Streaming Analytic Designs for Para.pdf](file://C:\Users\david\Zotero/storage/THKTVIBT/Najdataei - Efficient Data Streaming Analytic Designs for Para.pdf) file-path:: file://C:\Users\david\Zotero/storage/THKTVIBT/Najdataei - Efficient Data Streaming Analytic Designs for Para.pdf

• demand for timely analysis has resulted in a shift of data processing paradigms towards continuous, parallel, and multitier computing. ls-type:: annotation hl-page:: 5 hl-color:: purple id:: 63b9875f-e9fe-4d67-912c-5f0faa98360d
• analysis speed, precision, costs, and deterministic execution ls-type:: annotation hl-page:: 5 hl-color:: purple id:: 63b9876b-29fd-4210-93e7-7e6e99b63a51
• efficient continuous processing of streams of data in a decentralized and timely manner. ls-type:: annotation hl-page:: 5 hl-color:: purple id:: 63b98774-1343-48d1-8d0c-02248b5c3cdc
• ontinuous machine learning/data mining types of problems, appearing commonly in IoT applications, and in particular continuous clustering and monitoring, for which we present novel algorithms ls-type:: annotation hl-page:: 5 hl-color:: purple id:: 63b98788-2649-45a6-be7f-74bd20ffbbd1
• DRIVEN ls-type:: annotation hl-page:: 5 hl-color:: purple id:: 63b987b5-712f-4a26-a3ab-38b34d0712b1
• STRATUM ls-type:: annotation hl-page:: 5 hl-color:: purple id:: 63b987c4-6879-4f67-84ca-0203ef9a7470
• STRETCH ls-type:: annotation hl-page:: 5 hl-color:: purple id:: 63b987ca-5000-4518-91af-b475c85a61aa
• Driven: a framework for efficient data retrieval and clustering in vehicular networks ls-type:: annotation hl-page:: 9 hl-color:: purple id:: 63b9891f-88c6-4520-9e13-714838168cf1
• oday, we are in the midst of the fourth industrial revolution, a.k.a. Industry 4.0, that optimizes the computerization from recent decades with the use of Internet technologies to create smart production and smart services ls-type:: annotation hl-page:: 17 hl-color:: purple id:: 63b98953-bf92-4ca3-a9dc-486903c1b63a
• automated systems with integrated computational and physical capabilities ls-type:: annotation hl-page:: 17 hl-color:: purple id:: 63b98961-6251-4bba-abe6-1c138a381265
• high volume and high rate, ls-type:: annotation hl-page:: 17 hl-color:: purple id:: 63b9898f-559d-4b15-bc44-d0a56bd5ff6f
• cloud computing ls-type:: annotation hl-page:: 17 hl-color:: purple id:: 63b9899d-08e3-47f2-84a9-1688d5980f51
• dramatic escalation in data volumes and data rates pushes the limits of centralized data processing infrastructures ls-type:: annotation hl-page:: 17 hl-color:: purple id:: 63b989aa-c77a-431a-affd-09fc8cb4e435
• for deriving timely intelligence ls-type:: annotation hl-page:: 18 hl-color:: purple id:: 63b989ad-ee3c-4da0-a48e-70b2b39cd317
• the amount of data generated by each device, as well as the need to immediately extract knowledgeable information from such data, make it inefficient and impractical to send all the traffic to the cloud, doing the processing there, and then receive back the results. hl-page:: 18 ls-type:: annotation id:: 63b989cc-6342-45de-a273-65483500905c hl-color:: purple
• the inefficiency of traditional data analytics, has begun a new wave of scientific revolution and led to innovative tools such as continuous processing that analyses flows of data on the fly. hl-page:: 18 ls-type:: annotation id:: 63b98a16-d8ba-4415-87b3-88e9802c7957 hl-color:: purple
• enable convenient and on-demand access to a shared pool of resources ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b55-db03-494a-88d1-576830eb7762
• In IoT applications, the main motivation behind employing clouds, consisting of high-end servers, is to carry out heavy data analysis ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b62-6ca5-4490-ac1d-856317fa386e
• IoT devices are often equipped with reduced computational power, i.e. they are resource-constrained devices, and hence likely to be less efficient in performing heavy analysi ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b6d-347b-49bb-9361-63b7b8222ada
• loud computing ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b7b-2b9f-4f8f-ae98-ad89cdd59645
• challenges ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b7c-fc8d-4f38-bdaa-c76b4ffdf06a
• when dealing with Big Data, it might be impossible to send all data to the cloud without exhausting the available communication bandwidth ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b84-204d-4a72-a4a1-293352225aeb
• sending data to the cloud for analysis and getting back the results to make further decisions, could cause significant high latency which is not tolerable for certain applications requiring real-time processing. ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b8e-9ae7-40b8-84c7-10b2dce0869b
• high latency causes a drastic effect on power and energy consumption and influences the reliability of the system ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b97-fb26-45f6-847b-600eeb777791
• degradation in the Quality-of-Service (QoS). ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98b9b-19c3-4927-8e35-172121f3fbe4
• fog computing and edge computing ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98ba1-20a3-4346-818b-1e30ea54dd64
• procedures are being pushed down closer to where data originates. ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98ba9-7159-460a-a4cf-38056473eb63
• ntelligence and data processing is offloaded onto the edge devices (e.g. routers, switches, sensors and actuators.) without being sent to the cloud ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98bbc-e430-44a5-a51a-50f57a6f7780
• Edge computing is especially beneficial for applications which require ultra-low latency and real-time analysis ls-type:: annotation hl-page:: 19 hl-color:: purple id:: 63b98bc8-5398-4d21-8112-1a67bf8557cf
• [:span] ls-type:: annotation hl-page:: 20 hl-color:: purple id:: 63b98c01-8b4d-4f37-9346-f1d2c0fdcde3 hl-type:: area hl-stamp:: 1673104384786
• By bringing the intelligence away from the cloud, each fog node performs data processing closer to the IoT devices and consequently reduces the amount of data transported. A fog node also can be seen as a mini-cloud, which is located near the edge layer and the IoT devices connected to ls-type:: annotation hl-page:: 20 hl-color:: purple id:: 63b98c4b-d05a-443e-879d-7a7d4628e438
• the edge/fog/cloud architecture brings several benefits as it allows applications to distribute the intelligence and take advantage of a diverse range of computing resources and storing assets ls-type:: annotation hl-page:: 20 hl-color:: purple id:: 63b98c76-74e4-4db5-b675-73032e781ef1
• to design efficient algorithms and frameworks to process data in such 3-tier architecture we need to know how and where to process the data. ls-type:: annotation hl-page:: 20 hl-color:: purple id:: 63b9963c-7fa3-49c5-aefd-0edf30f50aea
• earn from data, provide data driven insights, and make decisions ls-type:: annotation hl-page:: 20 hl-color:: purple id:: 63b99801-dcb5-4a41-a672-6ecdce343930
• adapt the traditional ML algorithms to continuously process the flow of data ls-type:: annotation hl-page:: 21 hl-color:: purple id:: 63b9980f-c351-4357-b2dc-16b7f8cdaa36
• Continuous processing allows improvements in memory access patterns as well as enabling real-time monitoring which is a requirement for many application ls-type:: annotation hl-page:: 21 hl-color:: purple id:: 63b99816-7bb7-44df-8d71-cc4e32957d96
• continuous processing and querying the constant stream of data received from a temperature sensor, it is possible to raise an alarm once the temperature reaches a certain threshold ls-type:: annotation hl-page:: 21 hl-color:: purple id:: 63b9982e-fe65-4cb7-b89f-a9806267ff7f
• As data processing requirements grow, IoT applications demand utilization of the whole spectrum of devices in the computing continuum, from cloud data centers to edge systems and endpoint devices. ls-type:: annotation hl-page:: 21 hl-color:: purple id:: 63b99839-28c4-4049-8ded-391301efc886
• ardware-aware processing approache ls-type:: annotation hl-page:: 21 hl-color:: purple id:: 63b9983f-3b77-4a79-bae2-31af226f3745
• performance analysis ls-type:: annotation hl-page:: 21 hl-color:: purple id:: 63b9985c-b3e2-4582-97ed-ab85409425af
• decision process of where to compute applications ls-type:: annotation hl-page:: 21 hl-color:: purple id:: 63b99864-af7b-4dcd-81fe-2f8038030366
• neural network consisting of multiple layers and large data sets is a better option for cloud computing while a collaborative data filtering can be run on fog nodes. ls-type:: annotation hl-page:: 21 hl-color:: blue id:: 63b99873-26f0-4e2d-ad9a-f5b9addda8f9 hl-stamp:: 1673107574096
• dge computing is recommended for applications with soft execution time constraints to reduce the energy costs and carbon footprints ls-type:: annotation hl-page:: 21 hl-color:: blue id:: 63b9987c-3d40-46ae-8eff-e9aee9ef0481
• Figure 2 illustrates an example of smart transportation system that utilizes edge, fog, and cloud computing. As shown in the figure, the data regarding street conditions and unexpected events such as accidents are generated by many sensors deployed in the cars. This data, then, can be processed locally by resource constrained devices next to the sensors, to be used by the cars to adjust the velocity and avoid obstacles (edge computing). Cars can also send such data to the nearby traffic lights to be processed by the small embedded servers and extract information in order to adjust the lights, thus, relieve potential traffic congestion around the reported location (fog computing). Moreover, the city services can receive data from cars as well as the transportation infrastructures including traffic lights, to process the aggregated data for making further decisions (cloud computing ls-type:: annotation hl-page:: 21 hl-color:: blue id:: 63b998db-4004-4323-9087-af18f15fafd9
• [:span] ls-type:: annotation hl-page:: 22 hl-color:: blue id:: 63b998ec-5b05-4ea1-b75a-8d870292da65 hl-type:: area hl-stamp:: 1673107692010
• owards a computing continuum: Enabling edge-to-cloud integration for data-driven workflow ls-type:: annotation hl-page:: 47 hl-color:: blue id:: 63b99913-facc-47b4-a489-8b6ea38aae9f
• 22] D. Balouek-Thomert, E. G. Renart, A. R. Zamani et al., ‘Towards a computing continuum: Enabling edge-to-cloud integration for data-driven workflows,’ The International Journal of High Performance Computing Applications, vol. 33, no. 6, pp. 1159–1174, 2019 (cit. on p. 7). ls-type:: annotation hl-page:: 47 hl-color:: blue id:: 63b999ae-a400-44e6-a30e-b6f739b2ae64