Hierarchical Wireless Sensing And Identification Systems For Internet-of-Things

Internet-of-Things is a hot research topic in communications and networks. The 12th five-year plan of China has regarded Internet-of-Things as one of the strategic and emerg-ing industries. Around how to build hierarchical wireless sensing and identification sys-tems for Internet-of-Things, this dissertation discusses theoretical analysis, simulation me-thod, sensor node design, and system implementation. It focuses on the following four as-pects:(1) Protocol design on PHY layer and MAC layer for hierarchical RFID networks(a) Hierarchical multi-reader passive RFID networks are proposed to build the infrastructure of Internet-of-Things. We employ asymmetrical wireless links, using UHF as the forward link for reader-to-tag communication and TH-PPM IR-UWB as the backward link for tag-to-reader communication. We analyze the different parts of the system delay in the communication procedure, and propose relevant strategies to shorten the system de-lay. The length of TH sequence and the number of readers are optimized.(b) CDMA technologies are proposed in the backward link. Tag clock frequency is not increased, and a reader uses a single-user detection scheme and only detects the strongest tag signal. A new dynamic framed slotted ALOHA scheme and the relevant tag estimation algorithm are proposed to reduce tag-to-tag collisions. We analyze the system performance theoretically and optimize the length of Gold codes and the number of read-ers. According to the experiment results, our proposed protocol is more efficient than the EPC Class-1 Generation-2 protocol.(2) Simulation framework and method for Internet-of-Things(a) A new approach for simulating multiple wireless standards simultaneously within one case is proposed in OMNeT++. It is realized by partitioning and modeling the protocol stacks of different standards and designing a multi-radio module. We conduct a case study on this approach to simulate a two-layered wireless network based on IEEE 802.11 and IEEE 802.15.4a, and show its efficiency.(b) A CO-Simulation framework with MATLAB and OMNeT++(COSMO) is pro-posed. COSMO supports accurate indoor channel models and 3-D models, and has the abil-ity of self-validation. It combines the strengths of MATLAB and OMNeT++by compiling prebuilt models and powerful functions in MATLAB to header files and shared libraries and integrating them into OMNeT++. This avoids complicated synchronization technolo-gies between MATLAB and network simulator. Through a case study for simulating pas-sive RFID locating, COSMO can accelerate the design and reduce debugging time, and is more effective than current network simulators.(3) Sensor node design and networked servicesWe design and implement a fresh food tracking system for Internet-of-Things. The system consists of a two-layered wireless network. It adopts GSM/GPRS for the communi-cation between the server and master sensor nodes, and semi IR-UWB for the communica-tion between master sensor nodes and slave senor nodes. We develop the control platform on the server, implement the hardware design of master sensor nodes, and propose a prin-table model for slave sensor node. Cell-ID information in GSM network is adopted to track the entire logistic process. Field tests show the efficiency of the networked services and the reliability of the whole system.(4) Interpolation engine design for video sensing codecThe architecture and an application specific instruction set of a programmable inter-polation engine are proposed for video sensing codec applications in Internet-of-Things. It uses a very long instruction word (VLIW) and single instruction multiple data (SIMD) hy-brid architecture. Through the study of interpolation algorithms in different video stan-dards, we design an application specific instruction set. The interpolation algorithms in H.264 and AVS standards are realized with the proposed instruction set. We record the average execution cycles for each interpolation mode in test sequences, and realize the hardware implementation.