A new C-RAN architecture based on RF signal soft-switching

Abstract

Coordinated multi-point (CoMP) is known to be one of the key technologies for long-term evolution (LTE)-advanced systems. CoMP technology can improve system capacity and the quality of wireless communication services for users in LTE networks. However, in practice, the actual performance of CoMP technology is limited by the switching capacity of the backhaul network among distributed base stations as well as its latencies. In this paper, we propose a new cloud radio access network architecture based on RF signal soft-switching to solve this problem. Furthermore, we introduce a narrow-band parallel processing technique on a common public radio interface in downlink and uplink to reduce the volume of data as well as the latencies in the transmission process among base band units and remote radio units. By combining theoretical analysis with computer simulations, we show that the technique is valid both for downlink and uplink, i.e., it does not degrade the performance of downlink and uplink propagation between BBU pool and user equipments. Moreover, the computational time of the narrow-band parallel processing technique is less than that of the standard technique.

Appendix

We now prove that the noise caused by I/Q equalization on CPRI is actually an additional noise.

For an arbitrary base signal {x _i } which is generated and sampled in base station, it will be normalized, followed by I/Q equalization, before transmitted across CPRI. Let $\{x_{i}^{\prime }\}$ and $\{\bar {x}_{i}\}$ be the signals processed by normalization and equalization, respectively. We have

$$ x_{i}^{\prime}=\frac{x_{i}}{\max_{i}\{x_{i}\}}=\frac{x_{i}}{m} $$

(12)

and

$$ \bar{x}_{i}=\left\lfloor x_{i}^{\prime}(2^{Q-1}-1)+\frac{1}{2}\right\rfloor, $$

(13)

where Q is the number of bits of equalization.

After the base signal is received by RRU(s), it will be dequalized, followed by denormalization, obtaining original base signal. Let $\{\tilde {x}_{i}\}$ be the restored signal. Thus,

$$ \tilde{x}_{i} = m\cdot \frac{\bar{x}_{i}}{2^{Q-1}-1}. $$

(14)

Combine Eqs. 12, 13, and 14, we obtain

$$\begin{array}{@{}rcl@{}} \tilde{x}_{i} &=& m\cdot \frac{\left\lfloor \frac{x_{i}}{m}(2^{Q-1}-1)+\frac{1}{2}\right\rfloor}{2^{Q-1}-1}\\ &=& m\cdot \frac{ \frac{x_{i}}{m}(2^{Q-1}-1)+\frac{1}{2}-a_{i}}{2^{Q-1}-1}\\ &=& x_{i} + \frac{m(1/2-a_{i})}{2^{Q-1}-1}, \end{array} $$

where a _i is a number smaller than 1. Therefore,

$$\tilde{x}_{i}-x_{i}=\frac{m(1/2-a_{i})}{2^{Q-1}-1}=n_{i}. $$

We established the conclusion.