Abstract:         

Polar codes have been selected as the control channel coding scheme for 5G communications, because of their provable, capacity-achieving performance and low encoding/decoding complexity. Successive Cancellation (SC) decoding and Belief Propagation (BP) decoding are two approaches for decoding polar codes. SC is able to achieve good error-correcting performance, is less computationally expensive as compared to BP, but suffers from long latency due to its serial nature. To satisfy high throughput and low latency, BP is ideal as the decoding algorithm due to its inherent parallel processing nature. However, the error performance of BP is generally worse than that of SC, and since BP is iterative, the required latency and energy dissipation increases linearly with the number of iterations. Recently, many improved SC and BP based algorithms have been proposed, however, the suggested improvements often involve the use of large code lengths to observe exemplary error-rate performance. The increase in code length comes at the cost of increasing the hardware complexity. Traditionally, a method to reason between the practical feasibility of these algorithms, is to subjectively select a desired trade-off point between hardware complexity and error-rate performance. In our research, we propose using total energy analysis on a scalable, BP based unit, involving a sum of the operational hardware cost for a BP based decoding algorithm, and the cost of coherent RF transmission to achieve a required performance. With this analysis, we hope to introduce and establish a clear and objective evaluation criterion that reveals comparative practical feasibility of BP-based decoding algorithms of polar codes.