Queries In Heterogeneous Wireless Sensor Networks

Published Date: 02 Nov 2017

ABSTRACT

Demands on better interacting with physical world require an effective and efficient collaboration mechanism of multiple heterogeneous sensor networks. Previous works mainly focus on each single and specific sensor network, thus failing to address issues in the newly emerging scenario. In this paper, we propose COSE, a query-centric framework of collaborative heterogeneous sensor networks, where sensor networks collaborate with each other for effective and efficient processing of queries. Finding an optimal strategy of query processing with respect to energy efficiency is a crucial issue in COSE, which we formulate into an optimization problem, called EE-QPS. We prove the NP-hardness of EE-QPS, and then design a heuristic approach named IAP by utilizing the correlation (called implication in this paper) among different sensor networks. The experimental results demonstrate that in the context of COSE, IAP achieves optimized energy efficiency under various settings.

Index Terms

Sensor network, query processing, energy.

I.INTRODUCTION

Due to the recent advances in wireless communication and microelectronic technologies, both the price and size of sensors have decreased quickly. Today’s applications for sensor networks range from personal to mission critical systems including scientific observation, digital life, home automation, environment surveillance, traffic monitoring.Many of them are developed and promoted by governments, enterprises, and public organizations,offering continuous collection of real-time information,fulfilling the requirement of people’s daily lives.In the foreseeable future, we expect to witness the proliferation of sensor networks with a variety of functions that require a comprehensive collaboration mechanism among them.

Specific designs are necessary to manipulate a fabric of multiple sensor networks, facilitate the collaboration among them, and support efficient query processing. Previous studies in sensor networks, however, mainly focus on the performance and efficiency inside a single sensor network .In this work, we broaden the research into the scope of multiple sensor networks.

Previous studies mainly concentrate on data collection and query processing in a single sensor network. Using these approaches, we can only obtain isolated and incomplete results, inevitably leading to unilateral and even incorrect decisions. Also, the sensor networks continuously generate huge volumes of data with various attributes simply gathering all the data and processing them in a centralized manner is communication-intensive. Thus, distributed sensing and collaborative query processing among multiple sensor networks are indispensable for the above application. Moreover, the sensor networks are likely to receive substantive complex ad hoc queries, while the sensors are usually energy constrained and not easily rechargeable. Therefore, energy-efficient query processing with multiple sensor networks is a crucial issue but has never been studied before. Considering that all the sensor networks are spatially distributed and independent, how to enable them to effectively collaborate is a challenging issue, even if the optimal strategy of query processing is provided. To address the above challenges, we propose COSE, a query-centric ramework of collaborative heterogeneous sensor networks. Our main idea is to utilize the correlations

(called implication in this paper) among heterogeneous sensor networks to reduce the communication cost incurred by query processing and forwarding among the sensor networks.

The major contributions of this work are as follows:

1. We propose the framework of collaborative heterogeneous sensor networks and an implication-awarescheme called sink-overlay to organize heterogeneous sensor networks.

2. We formulate the optimization problem of query processing in COSE and prove its NP-hardness.

3. We design a mechanism to estimate the implications among sensor networks, based on which we propose efficient algorithms to schedule the pipeline of query processing.

II.PROBLEM DESCRIPTION

In heterogeneous sensor networks Previous works mainly focus on each single and specific sensor network, thus failing to address issues in the newly emerging scenario. The existing works focus on the issue of query processing in a single sensor network. They are similar with our scheme in the exploitation and utilization of correlations in query processing.

Drawbacks of Existing System:

Existing works focus on the issue of query processing in a single sensor network. This drawback will be overcome in proposed system by using heterogeneous wireless sensor networks.

III.PROBLEM SOLUTION

We propose COSE, a query-centric framework of collaborative heterogeneous sensor networks, where sensor networks collaborate with each other for effective and efficient processing of queries. A greedy algorithm is also proposed for faster decisions, which outputs the close-to-optimal scheduling of query processing. We then evaluate the proposed IAP approach with large-scale simulations.

IV.IMPLEMENTATION

We propose COSE, a query-centric framework of collaborative heterogeneous sensor networks. Our main idea is to utilize the correlations (called implication in this paper) among heterogeneous sensor networks to reduce the communication cost incurred by query processing and forwarding among the sensor networks.

COSE consists of a central manager (CM) and numerous collaborative heterogeneous sensor networks(sinks). Both CM and the sinks of sensor networks are connected over the Internet.

Central Manager (CM)

CM maintains a list of all active sensor networks in COSE.

The format of the information stored for each sensor network together with the formalization

It provide a uniform web portal that accepts external queries and outputs responses.

CM plays the role of scheduling query processing. It determines the pipeline of processing, and collects data from the queried sensor networks.

The Query-Centric Framework.

Sink Overlay:

The sinks serve as gateways for interconnecting the sensor networks in COSE.

Nodes Formalization

We provide a uniform formalization for sensor networks including the elementary properties, such as location, scale, function, energy, and data attributes, etc. This allows us to formalize a sensor network as a data source that produces meaningful data at quantified costs. We formalize a sensor network W by the following expressions. W is a sensor network in COSE containing four elementary domains W =<Basic;Energy; Data; Overlay>; where

Basic =<Name; Location; Scale>

Energy =<Unit cost;Current capacity>

Data =<Period of validity; Attribute W1;Attribute W2 . . .>

Overlay= <Neighbor List1;Neighbor List2 . . .>

Sink Overlay Construction

We construct the sink overlay using an implication-aware method. The overlay connections are categorized into two types, local connection and attributed connection, which are built among the sinks of the sensor networks. A local connection connects a sensor network with another one in its adjacent area. An attributed connection is built between two sensor networks that have common data attribute(s).

Sink-Overlay Diagram

Steps:

When a new sensor network joins COSE, it registers its information at CM.

Upon receiving the registration, CM checks its local database to returns a list of candidate neighbors to the new sensor network.

Each candidate neighbor has spatial proximity or(/and) functional proximity with the new one.

CM determines this list according to the information of the Basic and Data domains of the sensor networks.

After receiving the list, the new sensor network connects to all the neighbors in the list, either with local connections or attributed connections

Data Sharing Policy

Data sharing is carried out among neighboring sensor networks as a manner of measuring the implications among them. Besides on-demand sensing to resolve queries, every sensor network periodically collects its sensor data to track the status in the network. The sensor data from periodical and on-demand sensing are stored at the sinks until expiration. After a pre-configured period, every sensor network exchanges the latest sensor data with all its neighbors so as to keep them updated. Note that the frequency of periodical data sensing is configured much lower than that of ondemand sensing. Thus, the overhead of data sharing through

V.CONCLUSION

Considering that the future sensor network systems supporting numerous users and applications, it is an interesting and important issue to address multiple pipelines for query processing in heterogeneous sensor networks.Because the data of a sensor network may be shared/reused by multiple queries, the optimal schedule of a single pipeline does not match the optimal solution when it is scheduled together with other pipelines. Implication is no longer the only factor to be considered. High reusability of sensing data will build up the utility of a WSN in scheduling. Other issues, such as query’s priority and query aggregation are also potential research directions.

VI.REFERENCES

[1] T. Gao et al., "Participatory User Centered Design Techniques for a Large Scale Ad-Hoc Health Information System," Proc. First ACM SIGMOBILE Int’l Workshop Systems and Networking Support for Healthcare and Assisted Living Environments (HealthNet), 2007.

[2] R. Szewczyk et al., "An Analysis of a Large Scale Habitat Monitoring Application," Proc. ACM Second Int’l Conf. Embedded Networked Sensor Systems (SenSys), 2004.

[3] P. Zhang et al., "Hardware Design Experiences in ZebraNet," Proc. ACM Second Int’l Conf. Embedded Networked Sensor Systems(SenSys), 2004.

[4] Y. Liu et al., "Location, Localization, and Localizability," J. Computer Science and Technology, vol. 25, pp. 274-297, Mar. 2010.

[5] P.D.R. Fonseca, P. Levis, and I. Stoica, "Quanto: Tracking Energy in Networked Embedded Systems," Proc. Eighth USENIX Symp. Operating System Design and Implementation (OSDI), 2008.

[6] N. Xu et al., "A wireless Sensor Network for Structural Monitoring," Proc. ACM Second Int’l Conf. Embedded Networked Sensor Systems (SenSys), 2004.

[7] X. Mao et al., "Citysee: Urban CO2 Monitoring with Sensors,"Proc. IEEE INFOCOM, 2012.

[8] J. Hill and D. Culler, "Mica: A Wireless Platform for Deeply Embedded Networks," IEEE Micro, vol. 22, no. 6, pp. 12-24, Nov./ Dec. 2002.

[9] C. Intanagonwiwat et al., "Directed Diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks," Proc. ACM MobiCom, 2000.

[10] A. Deshpande et al., "Model-Driven Data Aquisition in Sensor Networks," Proc. 30th Int’l Conf. Very Large Data Bases (VLDB), 2004.

[11] K.A. Delin, "The Sensor Web: A Macro-Instrument for Coordinated Sensing," Sensors, vol. 2, pp. 270-285, 2002.

[12] P.B. Gibbons et al., "IrisNet: An Architecture for a World-Wide Sensor Web," IEEE Pervasive Computing, vol. 2, no. 4, pp. 22-33,Oct.-Dec. 2003.