Back to results list
Please use this identifier to cite or link to this item:
|Title:||Complex event detection in RFID and Wireless Sensor Networks|
|Keywords:||Radio frequency identification systems.|
Wireless sensor networks.
Hong Kong Polytechnic University -- Dissertations
|Publisher:||The Hong Kong Polytechnic University|
|Abstract:||Radio Frequency Identification (RFID) and Wireless Sensor Network (WSN) are important enabling techniques for data collection in many fields including mobile computing, pervasive computing, and internet of things. A great deal of useful information about surroundings is obtained through them, and then fed into upper-layer applications. Event detection is a data processing technique that is quite suitable for RFID and WSN applications. An event encapsulates raw data into a meaningful form that denotes a user-specified activity, and thus relieves the users from tedious underlying data processing. Compared with traditional event detection, new challenges are raised in RFID and WSN systems. The wireless communication used by them is error-prone and causes unreliable event detection results. Moreover, limited energy supply, computation capacity, memory and other resources in these systems demand more efficient and effective approaches of event detection. In this thesis, we investigate the event detection in RFID and WSNs to address these new challenges. We consider three important aspects of event detection: data collection, event aggregation, and event inference. First, we study data collection, i.e. collecting data from surroundings to applications. We focus on reliable data collection in mobile RFID systems. One unique characteristic of RFID is that usually multiple RFID tags communicate with one reader simultaneously, which may cause collisions and unsuccessful identification of all the tags. This problem is even more serious in mobile RFID systems since the tags are moving and timely identification is required. Specially designed anti-collision protocols are needed to improve the identification rate of RFID tags. We propose a schedule-based RFID anti-collision protocol which, given a high identification rate, achieves the maximal tag moving speed. The protocol schedules an optimal number of tags to compete for the channel according to their identification deadlines, so as to achieve the optimal identification performance. Simulation results show that this approach can increase the moving speed of tags by 120% compared with existing approaches, while achieving an identification rate of 99.999%.|
Second, we study event aggregation, i.e. merging several sub-events into a composite event, and eventually the user required events. We focus on two energy-efficient event aggregation problems in WSNs. One problem is optimizing event aggregation utilizing complex relations in an event. The other problem is optimizing the event aggregation involving multiple events with different latency constraints and event relations (denoted by aggregation function). For each user specified event, a routing tree called event aggregation tree is usually built in a WSN to enable information exchange among sensor nodes for event aggregation. For the first problem, we utilize the complex relations included in an event to optimize the event aggregation tree. We propose principles of designing such an event aggregation tree. After that, we propose centralized and distributed algorithms to build this tree to achieve energy-efficient event detection. For the second problem, we first propose an approach to build energy-efficient event aggregation tree for individual events considering both latency constraint and aggregation function. We further optimize the routing structure for the aggregation of multiple events to save energy, by making some events share event aggregation trees instead of building their own. For both problems, simulation results show that our algorithms outperform existing approaches and save a significant amount of energy. Third, we study event inference, i.e. infer the occurrence of an event through the information of other events. We focus on RFID reader localization, where detected RFID tags are used to infer the location of an RFID reader. It is a challenging task to achieve such an objective in the presence of long-lasting regional fault that means the RFID tags in a large region cannot response to the RFID reader for a long time period. We propose an effective localization approach which can tolerate such kind of fault, and define the quality index to measure the accuracy of a localization result obtained by our approach. Both 2D and 3D localization are discussed in our work. Our method also can be integrated into Multidimensional Scaling approach to solve network localization problem which involves multiple target objects to be located. We have taken extensive simulations and implemented an RFID-based localization system. In both cases, our approach outperforms existing approaches in localization accuracy and can provide additional useful quality information.
|Description:||xviii, 174 p. : ill. ; 30 cm.|
PolyU Library Call No.: [THS] LG51 .H577P COMP 2013 Zhu
|Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|b2616064x_link.htm||For PolyU Users||203 B||HTML||View/Open|
|b2616064x_ir.pdf||For All Users (Non-printable)||3.86 MB||Adobe PDF||View/Open|
Checked on Sep 18, 2017
Checked on Sep 18, 2017
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.