Fixed Beam Smart Antennas
Fixed-beam smart antennas are a simple but effective method to boost the downlink capacity of UMTS FDD. In this paper we compare the two possible strategies. Using a four element uniform linear antenna array, we ﬁnd the optimum sector dependent on the direction of departure (DoD) spread at the base station. For the ﬁrst method we ﬁnd the optimum number of beams to be four for low DoD spreads and two or three for large DoD spreads. For the second method the optimum number of beams per sector is seven for small DoD spreads and goes down to four or ﬁve beams per sector for large DoD spreads depending on base station spacing. By extensive system level simulations, we show, for 1km inter base station distance, a capacity gain of more than 160% over a conventional 3-sectored reference system by both ﬁxed beam methods.
THE ﬁrst Universal Mobile Telecommunication System (UMTS) networks are currently launched in Europe and Asia. These networks will bring high data rate services to the mobile user. As the number of subscribers increases there will be the need to extend the capacity of the initially deployed UMTS networks. Smart antennas are a possibility to increase the capacity of UMTS without the need of additional sites or additional spectrum. Smart antennas exploit the spatial domain of the mobile radio channel with the help of an antenna array and appropriate signal processing at the base station. In literature, there are several approaches for the implementation of such smart antennas. These approaches can be divided into two main strategies, the ﬁxed beam methods and methods that apply user speciﬁc beam forming. In ﬁxed beam methods are studied. In user speciﬁc beamforming is studied, while discuss both, ﬁxed and user speciﬁc beamforming. In this paper we focus on ﬁxed beam methods that lay a speciﬁc number of ﬁxed beams over the coverage area. All mobiles in the coverage area of one beam are served by this beam. Thus in the ﬁxed beam approach, typically more than one mobile is served by a speciﬁc beam. The beams that cover a sector can be produced either with a passive (analog) beamforming network or digitally in base band. Base band beamforming requires phase coherency all the way to the antenna elements, which is not needed if a passive network is used for beamforming For the use of ﬁxed beams in UMTS there are two possible strategies. The ﬁrst method is to use the beams to increase the sectorisation of a base station site, which is fully compliant to Release 99 and later versions of the UMTS speciﬁcation and needs no cooperation by the Radio Network Controller (RNC). The disadvantage of this method is that, due to the different scrambling codes used in the different beams, which we will call logic cells, the data transmitted on the different beams is not orthogonal and the interference in regions where two beams overlap is quite high. This is avoided in the second method, which we will call switched beam method, where the beams carry only user data and a Secondary Common Pilot Channel (S-CPICH) to improve the channel estimation in the mobiles. The data channels transmitted with the different beams are typically scrambled with the same scrambling code so that their orthogonality is preserved. Therefore overlapping of neighboring beams is not as detrimental as for the ﬁrst method. This method needs uplink measurements at the base station in order to determine the best beam for downlink transmission. For a ﬁxed antenna array size at the base station, the performance of these two systems in terms of number of served users at a certain average user satisfaction depends heavily on the number of ﬁxed beams per 120 ± sector and the allocated power for the pilot and common channels. Due to the nature of the two strategies the optimum number of beams per 120 sector will be different. We think that, for a fair comparison of the performance of the two strategies, it is essential that the two systems are compared at an operational point (number of beams per sector and pilot channel power setting) that is optimum for each method. To the knowledge of the authors there is only one paper that compares these two ﬁxed beam methods. But this comparison is done with the assumption of six ﬁxed beams per 120 ± sectors for both methods. These are, in fact, too many beams for the logic cell method and too few beams for the switched beam method as we will show in the following.
On the Optimum Number of Beams for Fixed Beam Smart Antennas in UMTS FDD