Yu Hanxiao

Centralized Spectrum Sharing Using Reinforcement Learning

Spectrum sharing technologies are a kind of important technologies to overcome the shortage of the wireless spectrum, among which the centralized spectrum sharing exhibits superior performance in dense region. However, this method does suffer from large computational complexity and impractical implementation when optimization target is and overall system is complex. In this paper, reinforcement learning based centralized spectrum sharing is studied. Simulations results show that the proposed method holds good performance.

A Novel Feedback Method to Enhance the Graphical Slotted ALOHA in M2M Communications

The multiple access issue caused by the massive connections of devices is the key design aspect in the machine-to-machine communication system. As an uncoordinated access scheme, coded slotted ALOHA (CSA) is proposed and well studied to enable random access and high throughput simultaneously with no grant process. It shows efficient performance when the payloads are small. However, the CSA does suffer from the stopping set problem: the Successive Interference Cancellation (SIC) decoding process of CSA would come to a jam when normalized offered traffic is large. We propose an enhanced scheme based on CSA, that is, adding a novel physical layer feedback scheme on CSA to initiate the SIC decoding when there is no degree-1 slot. Considering the overhead of the feedback, simulations show that the proposed scheme can increase the number of successfully accessed devices in one frame.The multiple access issue caused by the massive connections of devices is the key design aspect in the machine-to-machine communication system. As an uncoordinated access scheme, coded slotted ALOHA (CSA) is proposed and well studied to enable random access and high throughput simultaneously with no grant process. It shows efficient performance when the payloads are small. However, the CSA does suffer from the stopping set problem: the Successive Interference Cancellation (SIC) decoding process of CSA would come to a jam when normalized offered traffic is large. We propose an enhanced scheme based on CSA, that is, adding a novel physical layer feedback scheme on CSA to initiate the SIC decoding when there is no degree-1 slot. Considering the overhead of the feedback, simulations show that the proposed scheme can increase the number of successfully accessed devices in one frame.

A Hybrid Automatic Repeat reQuest Scheme Based on Maximum Distance Separable Codes

For communication systems, a good error control technology is expected to get higher data transmission rate without reducing the quality of service. This paper presents a Hybrid Automatic Repeat reQuest (HARQ) scheme based on Maximum Distance Separable (MDS) codes to improve the ability of error correction. We divide the file into several segments, i.e., information packets and get a check packet using a kind of MDS codes before we transmit the file. Then we try our best to recover the file with the help of the check packet and information packets at the receiver. It is shown that our proposed HARQ scheme has better Block Error Rate (BLER) performance when compared to the traditional HARQ scheme, but the average slots cost per file does not increase significantly.