Polar Code Design for Irregular Multidimensional Constellations

Polar codes, ever since their introduction, have been shown to be very effective for various wireless communication channels. This, together with their relatively low implementation complexity, has made them an attractive coding scheme for wireless communications. Polar codes have been extensively studied for use with binary-input symmetric memoryless channels but little is known about their effectiveness in other channels. In this paper, a novel methodology for designing multilevel polar codes that works effectively with arbitrary multidimensional constellations is presented. In order for this multilevel design to function, a novel set merging algorithm, able to label such constellations, is proposed. We then compare the error rate performance of our design with that of existing schemes and show that we were able to obtain unprecedented results in many cases over the previously known best techniques at relatively low decoding complexity.