• No se han encontrado resultados

Uplink

It has already been deduced that in order to support the new paradigm shift in 5G mobile communications, in providing significant capacity increase compared to any current cellular solutions [152], radically new solutions need to be developed for the air interface.

Cooperative scheduling has only recently been promoted as a possible solution for interference mitigation by supporting major cell densification which in turn can lead to required 10-fold increase in spectral efficiency [152] [153].

Investigations into new fibre-wireless (FiWi) networks and CoMP techniques such as adaptive antennas [154], to avoid inter-antenna communication and significantly enhance cooperative transmission have also emerged [43, 156].

In contrast to the above proposals a receiver beamforming technique is conceived through comparing and contrasting literature, implemented and evaluated, with a focus on interference mitigation.

As described in Figure 4-1(a), the network eNodeBs are connected to the central office via a high speed optical fibre infrastructure which has not been critical in the justification, performance evaluation and significance of the forthcoming ideas of this thesis but puts a mark on the inspirations for the network coming next. eNodeBs are connected using the X2 interface where inter base station communication becomes possible. A point to point

Target UE Interfering UE Interferin g UE 500m (ISD)

Cloud Radio Access Network (C-RAN) High-speed fibre connectivity High-speed back-haul eNodeB 2 eNodeB 1 eNodeB 3 eNodeB (a) (b) Core Network X2 Proposed receiver beamforming technique

Figure 4-1: (a) FiWi system architecture with (b) the proposed receiver beamforming technique

92 connectivity has been described however other topologies, such as tree based passive optical networks [43], are also highly possible. Inter-antenna communication for the cooperative transmission is established using the X2 interface thus reducing significantly the latency of the proposed algorithm, since it does not have to go through the central office [43].

Furthermore, Figure 4-1(a) demonstrates a typical cellular deployment with the inter-site distance (ISD) of 500m. However the proposed techniques are also applicable for smaller ISDs typically encountered in dense urban environments. As also described in the figure, user signals transmitted in uplink can reach neighbouring cells reducing significantly the signal- to-interference-noise ratio (SINR) of that cell.

The proposed receiver beamforming technique to minimise this effect is described in Figure 4-1(b). Compared to typical beamforming transmission techniques applied in downstream, the proposed receiver beamforming uses the antenna reciprocity properties to steer the antenna receiver pattern towards the intended user thus increasing the uplink SINR. The antenna reciprocity feature is adopted from the antenna theory stating that the antenna properties such as gain and radiation pattern are symmetrical for transmitting and receiving. Therefore, as shown in Figure 4-1(b), eNodeB adjusts the receiver radiation pattern towards the targeted user while nulling the pattern at the interfering user equipment. This means that the main beam is put in the direction of the desired signal while nulls are in the direction of the interference. Each eNodeB is aware of the overall distribution of all UEs across the network thus effectively scheduling the transmission without additional delays.

Therefore, exploiting the aforementioned antenna symmetry, the individual receiving antenna elements at the eNodeBs are weighted using complex weights to create the desired receiver antenna pattern. The weights are computed by an adaptive algorithm at the eNodeB. Depending on the calculated weights, the eNodeB then selects the most appropriate receiver

beam from the pre-defined beam pattern matrix. The receiver beams in the direction of the desired UEs are added constructively whereas they are nullified in the directions of interfering UEs (i.e. null steering) as described in Figure 4-1(b). Since the proposed technique uses Closed Loop Spatial Multiplexing (CLSM) [177] that relies on channel feedback, the eNodeB can inform UEs for the preferred transmit antenna based on an established receiver pattern.

Figure 4-2 [178] illustrates the main elements of the receiving part of a smart antenna. The antenna array contains of N (e.g. N = 4) elements. The smart antenna reception part consists of four units which include the antenna array, radio unit, beam forming (DoA) and signal processing unit [12].

94 Normally the array (antenna) will have a low number of elements in order to avoid unnecessarily high complexity when it comes to signal processing. The radio unit consists of analogue-to-digital converters (A/D) and also N down-conversion chains (digital down converters), one for each array element.

The adaptive algorithm/ signal processing unit will, calculate the complex weights (W1,…,WN) based on the received signal from the UEs, with which the received signal from

each of the array elements is multiplied. These weights will then give rise to the antenna gain pattern in the uplink direction. The weights could then be slowly adapted to steer the beam until maximum signal strength occurs. In beamforming, the weights are chosen to give a radiation pattern that maximizes the quality of the received signal. However in switched beam technique the weights are optimized to maximise the received signal from the desired user. In adaptive array technique it will suppress the signals from interference sources to maximise the SINR. With N antenna elements, smart antenna can “null out” N-1 interference sources [179], but due to multipath propagation this number is typically lower.

There are different methods for calculating the weights depending on the type of optimisation criterion. In the switched beam technique, the receiver tests all pre-defined weight vectors (corresponding to the beam set) and choose the one giving the strongest received signal level. If the adaptive approach is used, which directs the receiver beam to attain a maximum gain towards the strongest signal component, the DoA is first estimated and then the weights are calculated. As mentioned in Chapter 3 the DoA and AoA can be estimated by one of many methods. When the beam forming is done digitally (after the A/D), the DoA and adaptive algorithm units can be integrated normally in the same unit (Digital Signal Processor).

The algorithm proposed in this thesis is based on the novel concepts of switched and adaptive antenna receiver beamforming. In switched beam the receiver antenna beam patterns are

fixed in time and users are switching from one beam to another depending on their location within the cell. This technique is beneficial to deployment scenarios where channel conditions do not change frequently. Alternately, the receiver beam pattern can be adapted according to channel conditions and users locations. Therefore, enhanced performance with respect to SINR is expected with increased implementation complexity compared to switched beamforming. A hybrid approach is also possible and is further investigated, contributing one of the novelties of the research.

Documento similar