Set optimization for efficient interference alignment in heterogeneous networks

• Autor(a) principal: Castanheira, Daniel
• Data de Publicação: 2014
• Outros Autores: Silva, Adão, Gameiro, Atílio
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/12528
• topic: Small-cells; Interference Alignment; Zero-Forcing; MIMO Systems; Diversity Methods; Codebook Design; Rayleigh Channels; Feedback; Random Vector Quantization

To increase capacity and offload traffic from the current macro-cell cellular system operators are considering the deployment of small-cells. It is expected that both the small and macro-cells will coexist in the same spectrum resulting in unsustainable levels of interference. Interference alignment is considered as an effective method to deal with such interfer- ence. By using interference alignment the small-cells align their transmission along a common direction to allow the macro-cell receiver to completely remove it. It is clear that if the two systems have no limitations on the information that may be exchanged between them to perform the signal design, then the performance may be improved in comparison to the case of no or partial cooperation. However, this full cooperation strategy requires a high-rate connection between the macro and small-cells, which may not be available. To overcome this problem we consider that the alignment direction is selected from a finite set, known to both macro and small-cell terminals. We provide sufficient conditions for this set that guarantee full-diversity, at the macro- cell, and propose an efficient method to optimize the set elements. Results show that an alignment set with a description length of 1 bit is enough to achieve the same diversity as in the case where an infinite amount of information is exchanged between both systems. The proposed set optimization method achieves better performance than random vector quantization and similar performance to Grassmannian quantization.