Multihop Power Scheduling and MIMO Relay Channel Estimation
- Author(s): Kong, Ting
- Advisor(s): Hua, Yingbo
- et al.
While single-hop wireless networks are commonly deployed today, multihop wireless networks are still in early stages of development. These networks have tremendous potential to be the technology of choice for providing ubiquitous Internet connectivity; minimizing the need for expensive wired infrastructure; and are relatively easy to deploy and maintain. However, there are still many fundamental challenges in wireless multihop relay networks. In this dissertation, we address two challenges that are of significant importance for wireless multihop relay networks.
The first challenge we address is multihop transmission and power scheduling. In a multihop relay network, there are two types of nodes: terminal nodes and router nodes. Terminal nodes transmit directly to the local router node with a single hop. Router node collects the data from its local terminal nodes and is responsible for transmitting these data to the access point (AP). As the router nodes are more sophisticated than the terminal nodes, we assume they can support complex signal processing techniques such as Dirty Paper Coding (DPC) and can support multiple antennas communications as well. In this dissertation, our goal is to design a transmission scheme, which can balance the power consumption in each router node so as to prolong the network lifetime.
We first propose a DPC based mutihop transmission scheme. An optimization problem of power scheduling and rate allocation to minimize a power related objective function or to maximize a rate related objective function is formulated. Then, a general gradient projection method is proposed to solve the optimization problem for networks where both single antennas and multiple antennas can be equipped in each node. Some useful properties are explored to realize fast computation. Furthermore, an alternative subgroup method is also provided to reach a tradeoff between performance and complexity when the network size becomes large. Numerical results show that our proposed method achieves better power saving and power balance performances compared with existing schemes.
The second challenge we address is the channel estimation and training design for multihop relay channels. We consider a two-hop amplify and forward (AF) Multiple Input Multiple Output (MIMO) channel first. To overcome the ambiguity problem in channel estimates, we propose an innovative channel estimation scheme. This scheme has two phases. In the first phase, the source transmits no signal while the relay transmits and the destination receives. In the second phase, the source transmits, the relay amplifies and forwards, and the destination receives. At the destination, the data received in the first phase are used to estimate the relay-to-destination channel, and the data received in the second phase are used to estimate the source-to-relay channel. The linear minimum mean square error estimation (LMMSE) is used for channel estimation, which allows the use of prior knowledge of channel correlations. The algorithms for finding the optimal source training matrix used at the relay for the first phase, and the optimal source training matrix at the source and the optimal relay training matrix at the relay for the second phase, are developed. Power allocation along the diagonals of source and relay training matrices is solved by using an alternating algorithm with low complexity and fast convergence. The two-phase LMMSE based channel estimation method for two-hop AF MIMO relay channels can be extended for multihop AF MIMO relay channel estimation.
In summary, we discuss and provide solutions to two critical challenges in wireless multihop relay networks: multihop transmission and power scheduling; MIMO relay channel estimation and training design. Our work advance the state-of-the-art in wireless multihop relay networks, and bring us closer to realizing the vision of ubiquitous multihop relay networks.