Research On Power Allocation Optimization For Wireless Hybrid Digital-Analog Transmission

With the rapid development of mobile internet technology, the problems and challenges in traditional digital communication systems are becoming more and more serious. The source coding and channel coding are studied independently in traditional wireless digital communication systems based on the Shannon’s separated coding theorem. The result of source and channel coding is to transmit the data with an optimal rate under a target channel condition. However, the channel quality of a mobile user is always changing, resulting in the mismatch of channel quality and target transmitting rate. Thus, the mobile user in a wireless digital communication system may suffer the "cliff effect". The receive quality will saturate when the channel quality gets better due to the existence of quantization errors, which is called "level-off effect". On the contrary, the receive quality will decrease dramatically when the channel quality gets worse than a threshold, which is called "threshold effect". The "cliff effect" consists of "level-off effect" and "threshold effect". The "cliff effect" in wireless digital communication systems has a serious impact on uses’experience, especially for the real-time service such as video and audio.Though the "cliff effect" does not exist in analog communication systems, the transmission is very inefficient because of the absence of compression. In order to take both the advantages of digital and analog systems, the idea of hybrid digital and analog (HDA) transmission is proposed where the transmitting data is split into two parts:digital and analog parts. The transmitting signal is a hybrid mode of digital and analog signals. The digital part improves transmitting efficiency and the analog part improves the robustness to the channel quality.However, as a new communication scheme, there are many problems and challenges to be addressed for HDA transmission. One of the important issues that should be considered is the power allocation problem. No matter how the digital and analog parts are transmitted, they will compete for the channel resource. The allocation of resource between digital and analog parts has great influence on the system performance.In this paper, we address the power allocation problem for HDA transmission. We focus on the power allocation in two communication scenarios:the HDA transmission for the source data with different importance in wireless one-hop communication; and the HDA transmission in a wireless relay network (WRN). In the first scenario, the transmitting source data contains a series of chunks with different importance. Each chunk has a different effect on the end-to-end distortion, which means the difference of importance of each chunk should be considered when allocating the power among them. As for each chunk, the power allocation between digital and analog signals should be utilized.Considering the fact that the receiver may be too far away from the transmitter to directly communicate with each other in real communication environment, we expand the wireless one-hop communication to the scenario of WRN. In this scenario, a lot of researches have shown that the performance of a WRN can be improved by appropriately allocating the power between the source and relay. On the other hand, there are also a few researches focusing on HDA transmission in a WRN, which only consider the power allocation between the digital and analog signals. However, both the power allocations on the two dimensions have significant effects on the system performance. Therefore, they should be jointly considered.The main contributions of this paper can be summarized as the following aspects.1) The power allocation problem of HDA transmission for the source data with different importance is studied in a wireless one-hop communication system. The source data is modeled as a multi-Gaussian source with a series of Gaussian sub-source chunks with different variances. The value of variance indicates the importance of the chunk. Then, in order to minimizing the end-to-end distortion, the joint power allocation problem for HDA transmission is formulated as a programming problem, which is proven to be convex. An iterative algorithm is proposed to solve the problem. Simulation results show the validity of our proposed solution.2) We jointly considered the source-relay and digital-analog power allocation, which is, to our best knowledge, the first attempt to take these two dimensions into consideration. For the AF relay network, according our analysis, the joint power allocation problem can be addressed by solve a convex problem. The closed expression of the solution is shown in this paper. Simulation results show that compared with other power allocation schemes, our proposed schemes achieve considerable gains under various channel conditions.3) The joint power allocation problem of HDA transmission in a DF-based WRN is also studied in this paper. For the DF relay network, the joint power allocation problem is more complicated as it involves the decoding and re-encoding operation at the DF relay. We separate the problem into two scenarios according to the channel qualities of S-R and S-D. In each scenario, the problem is formulated into a nonlinear fractional programming problem, which is very difficult to be solved directly. We transform the object function into a polynomial by introducing a non-negative parameter, resulting into an equivalent programming problem. An iterative algorithm is proposed to search the optimal solution according to Dinkelbach’s Method. Simulations are evaluated to validate the effectiveness of the proposed solution.