Write an essay on Efficient Indexing and Query Processing of Model-View Sensor Data in the Cloud ?
Advanced technologies are currently being used in widely in various transportation systems, solely depending upon the research interests. The sensor data management systems are built by a large number of sensors: the sensors being dependent upon the time and value available from the traditional systems. It is well known fact that the utilization of traditional system leads to a number of problems including the difficulties in the transportation of data across different platforms, thus, the utilization of modern sensor networks are encouraged by the domain experts (Kang 2013). The research subject presented in this paper has been conducted to analyze the factors on how sensor data management system and the techniques and models are being used in the structural explanation of the use of Model-View Sensor Data. This research has focused light upon the structural explanation and critical analysis by observing, gathering data, comparing previous research works to show efficiency in the further period.
Through this research work, we intend to find answers to the following sets of questions:
1. How can the sensor data management systems increase the value offered by the information system to provide services to the consumers?
2. What are the techniques are being used to for the management of sensor data?
This research work is being conducted so as to meet the following objectives:
1. To conduct the detailed review of existing literary works base on the utilization of cloud based query processing and indexing of data generated by Model-View Sensors.
2. To conduct a detailed study on the various technologies that are used for the management of data collected from sensor networks.
In the case of the project under consideration, time series management had been an important research problem. The researchers have provided a detailed discussion on two models that are currently being used for the management of data generated by Model View Sensor networks. In the following section of the report, has been provided on the models that have been identified by the researcher.
Model View Sensor Data Management:
According to Guo et al. (2014), model view sensor data management models are focused on some specific techniques for processing data queries- sensor time series segmentation. Yu and Korkmaz (2015), on the other hand, are of the opinion that Model View Sensor Data Management systems utilize a type of algorithm that fragment time series into disjoint segments and satisfy a series of mathematical functions or models for the storage of key values. HBase cluster, which is essentially a table region, stores a sequential range dividing of the row key space (Sathe et al. 2013). According to Yu, Sen and Jeong (2013) interval index in HBase depends on the key value store, the time of interval, the value range and model formula that can fulfill one data segment. Such different possibilities occur while organizing row key and columns for storing and querying data. Yu and Korkmaz (2015) invented time value and interval or model through the utilization of sensor data. Rows can be found out on the row key in the row value store where time range or point queries know the point to stop and need to start the scanning from the beginning of the table each time (Guo et al. 2015). Pleisch and Birman (2008) informed that the same thing happens in the case of value intervals as in the case of row keys stands, where the value ranges or the row keys are used to sort the boundaries of the sensor data. According to Pleisch and Birman (2008), while using the model of sensor data, four types of fundamental options are used for the purpose of querying: these include Time Point Query, Value Point Query, Time Range Query, Values Range Query.
Model of KVI-index (Key Value Interval Index ):
According to Keogh et al. (2001), KVI index consists of specific value range and time for the indexing purpose, in contrast to the process of indexing the mathematical functions of the time segments. KVI index is structured with a model index table and in-memory. The in-memory structure consists of a searching tree that operates virtually and is known as the vs-tree. It is a standardized binary search tree and a model index table is used to hold the key value store (Wang et al. 2013). The key value store generally is designed to be materialized with the secondary structures present in the nodes of the vs-tree (Rushinek and Rushinek 2015).
Comparing the Querying View Model:
Four basic standardized approaches have been reported that can be used for querying the data generated by the model view sensor. These approaches include MapReduce (MR), Interval Tree(IT), MapReduce+KVI (MRK) and Filter of Key Value Sore (FKV). These approaches depend upon the range query and point query.
Analytical view of KVI-MapReduce (MR):
MapReduce is based on a model that filters and grids with the effect of searching, finds depth as a means of the ratio of registered mode searching depth over the height of vs-tree in isearch+ (Guo et al. 2014).
Rushinek and Rushinek (2015) have proposed a model named the Innovative Interval Index. This model has been proposed for Model View Sensor data management in a key-value stores referring as in the case of KVI- index (Acharya and Mark 2013). This composite index frame can interestingly accommodate the new sensor data segment in an effective manner.
Cayirci et al. (2006) have, after inventing the KVI-index memory structure, introduced a new query processing approach known as the Hybrid model view query processing technique (Janowicz et al. 2013). This particular technique helps to integrate the range that can scan and process the segments accordingly.
An algorithm (isearch+) search has been generated by Pleisch and Birman (2008) that gives consecutive results by intersection search.
Yu and Korkmaz (2015) have introduced a framework that has been fully executed. The framework includes the online sensor data segmentation, KVI-indexing, hybrid data processing and modeling. Besides this, it has been compared with a large number of alternative techniques that are available and has been found to be of comparable efficiency.
Research Process and Outcomes
The entire research work that has been in this paper has been conducted by reviewing existing literary works and research articles that are based on the said research topic. The various models that have been proposed in the cited literary works have been studied diligently, and the information available from these data source has been analyzed.
Through the above mentioned process we have been able to identify the following processes that can be used effectively for the management of data collected through sensor networks:
1. Key Value Interval Indexing
2. Querying View Model
The utilities of these models have been provided in section 4 of this paper. Analyzing the data available from the cited literary articles and research papers has been helpful in reaching the conclusion that the efficiency of both the models are comparable when it comes to the management of data acquired through modern sensor networks.
In this project, the researcher has evaluated the key-value representation of an interval index for model view based sensor data management. It is different from the traditional external memory index framework in terms of complex mode merging and dual mechanism. The KVI index relies on two factors: the memory used to store the data and the materialized key value store. It helps to maintain the dynamic structure of sensor data generation in an easy manner. However, it has been analyzed that a Hybrid Query Processing approach named as KVI MapReduce. It contains KVI-index within a key value store. This research study has done extensive experiments on a raw database. The approach out-performs the former approach in terms of time and indexing efficiency.
Limitation and constraints
The entire research work has been based on data collected from exiting literary works: the authenticity of the secondary data being questionable. Therefore, specification of information is the major limitation in this study.
Besides this, the entire research work was highly constrained in terms of the time and financial resources. The less tightly scheduled research work would have been more effective in collecting authentic information and arriving at effective conclusions. Apart from that, in terms of maintaining ethical rules and regulations, analysts have to avoid some criteria that is also a potential limitations for this study work.
The research works reported in the reviewed literature have been of much help to the researcher in terms of comparing the efficiencies of the techniques used for the management of data collected through modern sensor networks. In the future, the research works can be conducted with the aim of developing a hybrid of the reported techniques. The objective of such research work being the development of a system with increased efficiency in terms of processing time and value retrieval by effective querying on the KVI-index. Since the absence of literary work on such topics indicates the lack of research works being conducted in this direction, it is expected that future works in this particular domain would facilitate the development of sensor data management system of higher efficiency.
Acharya, R. and Mark, D., 2014. Metadata Model, Resource Discovery, and Querying on large-scale Multidimensional Datasets.
Guo, T., Papaioannou, T. and Aberer, K. 2014. Efficient Indexing and Query Processing of Model-View Sensor Data in the Cloud. Big Data Research, 1, pp.52-65.
Janowicz, K., Bröring, A., Stasch, C., Schade, S., Everding, T. and Llaves, A., 2013. A restful proxy and data model for linked sensor data. International Journal of Digital Earth, 6(3), pp.233-254.
Kang, D. 2013. ‘Lightweight and Scalable Intrusion Trace Classification Using Interelement Dependency Models Suitable for Wireless Sensor Network Environment.’ International Journal of Distributed Sensor Networks, 2013, pp.1-10.
Pleisch, S., and Birman, K. 2008. Scalable querying of sensor networks from mobile platforms using tracking-style queries. IJSNET, 3(4), p.266.
Rushinek, A. and Rushinek, S., 2015. End-user satisfaction of data base management systems: An empirical assessment of mainframe, mini and micro-computer-based systems using an interactive model. ACM SIGMIS Database, 17(2), pp.17-27.
Sathe, S., Papaioannou, T.G., Jeung, H. and Aberer, K., 2013. A survey of model-based sensor data acquisition and management. In Managing and Mining Sensor Data (pp. 9-50). Springer US.
Wang, L., Tao, J., Ranjan, R., Marten, H., Streit, A., Chen, J. and Chen, D., 2013. G-Hadoop: MapReduce across distributed data centers for data-intensive computing. Future Generation Computer Systems, 29(3), pp.739-750.
Yu, B., Sen, R. and Jeong, D.H., 2013. An integrated framework for managing sensor data uncertainty using cloud computing. Information Systems, 38(8), pp.1252-1268.
Yu, X. and Korkmaz, T. 2015. Hypergraph querying using structural indexing and layer-related-closure verification. Knowl Inf Syst.