Study Of Resource Allocation In Broadband Over Power-line Systems

In the broadband over power-line of low voltage power grid (LVPG), it could effectively eliminate the effect of multipath, attenuation, time-variety and frequency-selection for power-line channel by applying the technologies of adaptive orthogonal frequency division multiplexing and dynamic resource optimization allocation (DROA). Based on the impressive research and survey on the development of DROA for power-line communications (PLC) systems, the study of this dissertation was focused on the application of DROA technology in the high-speed communications over the LVPG, and some strategies and algorithms of DROA based on the low-voltage power-line channel were proposed, which could better improve the PLC systems'utilization of resource and quality of service.Firstly, a bit-loading algorithm for rate adaptation based on the comparison of water-filling line and a resource pre-allocation and adjustment algorithm for power adaptation based on the optimal allocation rule were proposed for the single-user PLC systems with the single-phase power-line channel's characteristics, real restrictions and user requirements. Then they were tested in some typical power-line channel environment respectively and the simulation results indicate that they almost have the optimal performances on theirself resource allocation aims with low implementation complexity.Secondly, a resource pre-allocation and adjustment optimization algorithm for the mixture of power adaptation and rate adaptation was proposed for the multi-user single-phase PLC systems based on the differences of user's channel characteristics, and an algorithm of resource assignment for the first OFDM symbol and resource adjustment for the subsequent OFDM symbol in each slot-time based on the optimal allocation rule was proposed with the characteristic of power-line channel between multiple OFDM symbols in each slot-time, and a resource optimization allocation algorithm based on the resource factor was proposed with the multi-hiberarchy and multi-objective characteristics for the adaptive multi-user bit-power allocation on the multiple subcarriers, and an improved NSGA-â…¡genetic algorithm for single-and multiple-objective optimization with the mixture of power adaptation and rate adaptation was proposed with the globally searching characteristic of genetic algorithm, and a cross-layer multi-objective algorithm with the user-scheduling in data link control (DLC) layer and the resource allocation in physical layer was proposed, which selected the scheduling users from all users and ascertained their optimal cross-layer parameters in DLC layer and then given them to the resource allocation in physical layer, and which assigned the system's resource for the scheduling users in physical layer and then given the results of resource allocation to the next user-scheduling in DLC layer. The whole proposed algorithms were tested in some typical multi-user power-line channel scenarios, and the simulation results illustrate that they all could meet the multi-objective requirements of the multi-service resource dynamic allocation in PLC systems. Wherein, the performances of improved genetic algorithm are better than the general algorithms' and less than the resource factor algorithms'slightly.Finally, a bit-adding/subtracting table-looking algorithm based on the geometric mean decomposition decoupling the communication channels and the dirty-paper coding technology eliminating the directional interfere among the equivalent channels was proposed with the cross-interfere between the adjacent channels for the multi-conductor power cable communication systems, and another bit-adding table-looking algorithm based on the singular value decomposition decoupling the communication channels is also proposed for comparison. The proposed algorithms were tested in the various parameters of multi-conductor power cable and the various restrictions of PLC systems, and the simulation results indicate that the bit-adding/subtracting algorithm could keep the optimal system performances and decrease the complexity of resource allocation effectively.