Interference Modulation Order Detection with Supervised Learning for LTE Interference Cancellation

Blind detection of interference modulation order is studied in this paper. Exploiting the additivity property of cumulants for independent variables, we extend the techniques used in source automatic modulation classification to identify the interference modulation order. Using multi- class support vector machines, we show that accurate prediction performance can be achieved via supervised learning techniques. Three interference scenarios are considered - intra-cell interference, inter-cell interference with and without accurate interference channel estimates. Using numerical evaluation, we show that the proposed technique can be applied for all three scenarios without much degradation in all three cases. Furthermore, the technique is fairly invariant to changes in Signal-to-Noise Ratio (SNR) and Interference-to-Signal Ratio (ISR). These results are useful in advanced interference cancellation applications in wireless cellular communications, where the source modulation order is explicitly signaled, but not the interference.