5G connectivity has become essential to integrate rural communities into the broader digital economy and support critical applications like remote education and remote surgery. A major hindrance to expanding rural broadband coverage, especially in developing countries, is the high cost of installing 5G base stations. Hence, there is a need to reduce the cost of a 5G base station without degrading its performance. Our work proposes a novel approach to efficiently utilize the polar code encoders in a 5G base station. The idea is to use the idle time of the polar encoders during downlink transmission for error correction in the 5G data plane. Polar codes have conventionally been used in the 5G control plane, while LDPC codes are used in the data plane. We perform detailed characterization experiments to show the advantages of using polar codes in the data plane as well. Further, to intelligently distribute the user data packets among the available compute nodes, we propose a set of novel resource allocation algorithms and compare their performance with other algorithms in the literature. Using our proposed optimization techniques, we achieve a 17% reduction in the cost of a 5G base station. Simultaneously, we are able to improve the performance by 24% compared to a conventional base station.