Theses: Cooperative MIMO Communications - Information Theoretical Limits and Practical Coding StrategiesS. Simoens
THE RELAY-ASSISTED or COOPERATIVE transmission is a relatively new class of spatial diversity technique where a new element comes up in the conventional source-destination or point-to-point communication: an assisting relay or cooperating user. The relay assists the source in transmitting a message to the destination and allows dealing with the channel impairments like shadowing, multipath fading, interference and pathloss. Although the seminal works were issued in the 70's by van der Meulen, Cover and El Gamal, it has been during the last years when it has re-gained more attention by the researchers. In fact, the relay-assisted transmission can be seen as a virtual MIMO (Multiple Input Multiple Output) with distributed antennas. In contrast to MIMO systems, the transmission requires the use of additional channel resources because of the limitation of the current radio technology: the relay terminal is constrained to work in half-duplex mode, which motivates that the transmission must be carried out in two orthogonal phases (relay-receive and relay-transmit phase), duplexed in time or frequency domains.
This dissertation investigates protocols and strategies for the relay-assisted transmission which improve the spectral efficiency and homogenize the bit rate service in the cellular communication systems with uniformly distributed users. The new element present in the communication, the relay terminal, imposes a redefinition of many techniques and protocols commonly used in the point-to-point and MIMO systems, which are placed at the physical and upper layers.
First, achievable rates using the relay-assisted transmission are provided which depend on the role of the relay (amplify-and-forward or decode-and-forward), the type of the transmission (persistent transmission, incremental or selective relaying), the data transmitted by the relays (repetition or unconstrained coding) and the type of half-duplex protocol. There are up to four protocol definition depending on the activity of the terminals on each phase. An additional aspect addressed is the resource allocation for each phase, that is, either it is fixed beforehand (static) or it is adjusted dynamically (dynamic) as a function of the channel quality. For the single-user relay-assisted transmission the resources can be allocated based on the channel quality of the different links. Moreover, if there is complete channel state information about all channel coefficients (including the carrier phase of the transmitting terminals), source and relay can transmit synchronously enhancing the transmission thanks to the (distributed) eigenvector precoding.
Two relay-assisted transmission techniques are evaluated when a destination is assisted by multiple relays. Both depend on the messages intended to each assisting relay (independent or common messaging). The resource allocation for both techniques is shown to be convex.
Additionally, three different scenarios illustrate the multi-user relay-assisted transmission with a single destination and different types of half-duplex relays: RMAC (Relay-assisted Multiple Access Channel), UC (User Cooperation) and MARC (Multiple Access Relay Channel). The relay-assisted transmission can be done synchronously or asynchronously. The sources and relays are power limited and access in each phase of the communication by TDMA (time division multiple access), FDMA (frequency division multiple access) or SC (superposition coding multiple access). For those scenarios the allocation of transmitted power and time resources can be formulated as a convex problem under some circumstances, evaluating the optimal solution.
Afterwards, the relay-assisted transmission duplexed in time is applied to a centralized cellular system based on TDMA in the downlink. The reuse of one time slot for the transmissions done from the relays to destinations (relay slot) is proposed to improve the spectral efficiency. This solution produces interference for all the destinations active in that time slot. A power control algorithm (at the relays) based on game theory is proposed to combat the generated interference. Under that configuration a scheduler algorithm explores the multi-user gain for the relay-assisted transmission, measuring the introduced overhead.
Another way of dealing with the interference is by rate control management. Under some circumstances it is possible to model the probability density function (pdf) of the interfering power. In such a case, the source can tune the transmission rate in order to maximize the throughput. This solution is extended to the case where each destination is assisted by multiple relays. In spite of the interfering power, both proposed solutions are able to provide significant gains over the direct transmission.
Finally, the dynamic link control of the relay-assisted transmission is investigated under two different assumptions on the knowledge about the channel: statistical knowledge of the channel state and actual information about the current channel state. Both types of knowledge lead to different transmission strategies, in terms of selecting the modulation and coding scheme (MCS). Under the first case, the transmission rates are not adapted to the current channel realization and the destination can decode wrongly the messages. The Automatic Repeat reQuest (ARQ) protocols are redefined for the relay-assisted transmission to cope with these events. In this work we specify the (distributed) space-time codes, the coding at the source and relay and the length of the retransmissions. When there is actual information about the channel state the MCS can be adapted to the current channel realization. In such a case, the link error prediction for the relay-assisted transmission is investigated, and thus the MCS can be designed for maximizing the information rate for a given probability of packet loss or maximizing the throughput.