CLIAMS:1. A computer implemented method for data partitioning in an internet-of-things (IoT) network (100), the computer implemented method comprising:
determining, by a processor (110), number of computing nodes (106) in the IoT network (100) capable of contributing in processing of a data set (136);
ascertaining, by the processor (110), at least one capacity parameter associated with each of the computing nodes (106) in the IoT network (100) and with each communication link (108) between a computing node (106) and a data analytics system (102), the capacity parameter being indicative of a computational capacity for each of the computing nodes (106) and communication capacity for each communication link (108);
determining, by the processor (110), an availability status of each of the computing nodes (106) and each communication link (108), wherein the availability status is indicative of temporal availability of each of the computing nodes (106) and each communication link (108); and
partitioning, by the processor (110), the data set (136) into subsets, based on the number of computing nodes (106), the capacity parameter associated with each computing node (106) and each communication link (108), and the availability status of each computing node (106) and each communication link (108), for parallel processing of the subsets.
2. The computer implemented method as claimed in claim 1, wherein the at least one capacity parameter associated with each computing node (106) comprises processing speed of the computing node (106), available memory and associated bus speed, cache size, and operational load on the computing node (106) at a given point in time.
3. The computer implemented method as claimed in claim 1, wherein the at least one capacity parameter associated with each communication link (108) comprises data transfer rates for the communication link (108).
4. The computer implemented method as claimed in claim 1, wherein the determining the availability status comprises is at least one of advertisement-based, event-based, and poll-based.
5. The computer implemented method as claimed in claim 1, wherein the determining the availability status comprises ascertaining availability based on historical availability data.
6. The computer implemented method as claimed in claim 1, further comprising scheduling, by the processor (110), data processing tasks to each of the computing nodes (106) based on the partitioning.
7. The computer implemented method as claimed in claim 1, wherein the ascertaining the at least one capacity parameter associated with each communication link (108) between a plurality of computing nodes (106).
8. The computer implemented method as claimed in claim, 1, wherein the partitioning is at least one of time reduction-based and cost reduction-based.
9. A data analytics system (102) for data partitioning in an internet-of-things (IoT) network (100), the data analytics system (102) comprising:
a processor (110);
a capacity module (118) coupled to the processor (110) to,
determine a number of computing nodes (106) in the IoT network (100) capable of contributing in processing of a data set (136); and
ascertain at least one capacity parameter associated with each of the computing nodes (106) in the IoT network (100) and with each communication link (108) between a computing node (106) and the data analytics system (102), the capacity parameter being indicative of a computational capacity for each of the computing nodes (106) and communication capacity for each communication link (108);
an availability module (120) coupled to the processor (110) to determine an availability status of each of the computing nodes (106) and each communication link (108), wherein the availability status is indicative of temporal availability of each of the computing nodes (106); and
a partitioning-scheduling module (122) coupled to the processor (110) to partition the data set (136) into subsets, based on the number of computing nodes, the capacity parameter associated with each computing node (106) and each communication link (108), and the availability status of each computing node (106) and each communication link (108), for parallel processing of the subsets.
10. The data analytics system (102) as claimed in claim 9, wherein the availability module (120) determines the availability on at least one of event-basis, advertisement-basis, and poll-basis.
11. The data analytics system (102) as claimed in claim 9, wherein the availability module (120) determines the availability based on statistical analysis of historical availability data.
12. The data analytics system (102) as claimed in claim 9, wherein the partitioning-scheduling module (122) schedules data processing tasks to each of the computing nodes (106) based on the partitioning.
13. The data analytics system (102) as claimed in claim 12, wherein the partitioning-scheduling module (122) schedules data processing tasks based on static scheduling techniques.
14. The data analytics system (102) as claimed in claim 9, wherein the partitioning-scheduling module (122) achieves at least one of time reduction-based partitioning and cost reduction-based partitioning of the data set (136).
15. A non-transitory computer-readable medium having embodied thereon a computer program for executing a method for data partitioning in an internet-of-things (IoT) network (100), the method comprising:
determining a number of computing nodes (106) in the IoT network (100) capable of contributing in processing of a data set (136);
determining at least one capacity parameter associated with each of the computing nodes (106) in the IoT network (100) and with each communication link (108) between a computing node (106) and a data analytics system (102), the capacity parameter being indicative of a computational capacity for each of the computing nodes (106) and communication capacity for each communication link (108);
ascertaining an availability status of each of the computing nodes (106) and each communication link (108), wherein the availability status is indicative of temporal availability of each of the computing nodes (106) and each communication link (108);
partitioning the data set (136) into subsets, based on the number of computing nodes, the capacity parameter associated with each computing node (106) and each communication link (108), and the availability status of each computing node (106) and each communication link (108), for parallel processing of the subsets; and
scheduling data processing tasks to each of the computing nodes (106) based on the partitioning.
,TagSPECI:As Attached