Research On Resource Allocation And Scheduling For Perceptual Layer Of Internet Of Things

With the development of "Internet of Things" in recent years, more and more attention has been focused on it. Internet of Things (IoT) refers to a network of real-world objects linked by the Internet that connects each other through on-line services. It connects all the objects through kinds of sensing devices according to the agreed protocol, enables the exchanges of information and communication in order to realize the intelligent identification, positioning, tracking and monitoring. As the foundation and core part of IoT, the perceptual layer focuses on the information acquisition and identification. Then the research on how to achieve efficient resource allocation and scheduling for IoT perceptual layer has received widespread attention from academic and industry.RFID (Radio Frequency Identification) systems and WSNs (Wireless Sensor Networks) are two typical systems for the IoT perceptual layer. The paper focuses on the channel access control technology and sleep scheduling technology for the allocation of resources in the MAC layer. The main contents of this paper include the following aspects:1. Matching grouping dynamic frame slotted ALOHA algorithm based on Bayesian estimation for anti-collision is proposed. In dynamic frame slotted ALOHA algorithm, the estimation of the number of tags is the key factor that influence the algorithm performance. The traditional estimation algorithms rely solely on the current observation, which we call posteriori information, according to the classical statistical point of view. The estimation results depend on the number of trials times. With the problem in hand, the paper proposes an estimation algorithm based on Bayesian decision, which only needs a small amount of observational results and combines with prior information and posteriori information to get the accurate estimation. The suggested diagnosis method is deduced from theoretical analysis and simulations are made to verify the improvement of the reliability of estimation of the algorithm. Based on the accurate estimation, a matching grouping dynamic frame slotted ALOHA algorithm is proposed for the problem that the tag identification rate would decrease extremly with the number of tags rapidly increasing. It contributes the optimization grouping model by the mathematical theory, so that there are only a certain number of tags response the read in each identification cycle. Therefore, the probability of tag collision reduces and the system throughput improves. At last, the simulations also verify the effectiveness and stability of the algorithm.2. An anti-collision protocol called code division cooperative identification protocol (CDCIP) is proposed in order to tackle the reader collision problem in smart RFID systems. Traditional reading makes each reader work at the allocated time slots while keeping in "silent" state for the other time slots due to the interferences between the readers. As a result, the efficiency of the system is reduced. CDCIP allocates orthogonal sequences to each reader and composes a RFID network. Consequently, it strives to let more than one reader work at the same time and at the same frequency. Eventually, comparing with the traditional protocol, the simulation results show that the proposed protocol offers better average throughput and lower power consumption of the system.3. Spectrum-driven sleep scheduling algorithm based on reliable theory is proposed. Sleep scheduling mechanism is used in the tradition sensor network to conserve battery power and extend the network lifetime due to the limitation of node energy resource. In the new scene, cognitive radio sensor networks, the spectrum resources also become limited and need to be scheduled and allocated in addition to the energy resources. As a result, when the traditional sleep scheduling algorithms are used there will be two unavoidable situations, first is that some nodes in work state could not transfer the data because of lack of the spectrum resources. Secondly, some nodes access the spectrum resources but they are in the sleep state. Both of above are a waste of energy resources and spectrum resources. This paper introduces the reliability theory, and measures the spectrum resources with the indexes such as the failure frequency, the system availability and reliability function. On the basis of this, we realize optimal spectrum allocation and the corresponding sleep scheduling algorithm. This algorithm predicts the reliability of the spectrum hold time and the sleep time. The advantages of the algorithm are that it will ensure both the spectrum efficiency and the appropriate sleep settings to save energy and extend the network lifetime. Computer simulations show that the novel algorithms are effective in improving spectral efficiency and network lifetime.The main contribution of this paper is that, I perform the research of resource allocation and scheduling in three typical scenarios of two typical systems for IoT perceptual layer, RFID systems and wireless sensor network. For RFID tag collision scenario, Bayesian decision is introduced to estimate the number of tags and the tags are grouped by the optimal criteria to response the reader. In the scenario of RFID network reader collision, the algorithm is a CDMA-based centralized reader anti-collision protocol to make more than one reader work collaboratively and in parallel to improve system efficiency. For the sleep scheduling scenario in wireless sense network, reliability theory is introduced to solve the problem of mismatch between spectrum resources and the sleep cycle. It improves the spectrum efficiency and reduces the energy consumption of the system. All these algorithms achieve the optimal allocation of resources, improve the efficiency of the system in each scenaria and save the energy.