1536-1225 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LAWP.2016.2570818, IEEE Antennas and Wireless Propagation Letters > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1 Abstract — This letter deals with the problem of determining the number of RF sources faced by wireless communication systems. In presence of several sources it is unavoidable to have signal superposition, loss of tracking, and degrading (ambiguous) estimation. Moreover, the well-known subspace-based methods for tracking and estimating sources require prior knowledge of the number of sources. It is shown here that the Independent Component Analysis (ICA) method can be used not only to determine the number of sources, but also to increase the number of sensors (measures) by combining the array elements. The results also show that narrowband sources having a very small frequency difference (F = 0.41%) are detected. The ICA ability of blind sources separation makes it suitable to determine the number of unknown sources, which is crucial in subspace-based techniques. For instance, the ICA can be used as preprocessing model order estimation in Direction of Arrival (DOA) algorithms for smart antenna systems. Index Terms — Independent component analysis, FastICA, smart antennas, blind separation, DOA, location sensing. I. INTRODUCTION ANY widely used hand held devices and wearable sensors use the same operating frequency band. Intended and unintended radiated emission sources also run in the same band. Therefore, the receiving signals are mixed signals (sources and interferences), and, consequently, it demands for sophisticated methods to be estimated [1, 2]. This issue impacts the performance of radar, image processing, and communication systems. These are just a few examples of where adaptive antennas can be used, which has attracted much interest in the last decades. The efficacy of the estimation/location subspace-based methods such as SPIRIT and MUSIC begin with a prior knowledge of the number of sources, which is not a practical case. Therefore, the knowledge of the number of signals impinging on the array is crucial for those Direction of Arrival (DOA) estimation algorithms [2, 3]. In smart antenna technology, the adaptive processor consists of several computing processes [1]. The model order Manuscript received August 1, 2015. This work was supported by Brazilian Funding Agency CAPES under program CAPES/PROCAD. G. Fontgalland is with the Electrical Engineering Department - UFCG, Campina Grande, PB 58429-900 BR. (phone: +5583-2101-1721; fax: +5583- 2101-1418; e-mail: fontgalland@dee.ufcg.edu.br). P. I. L. Ferreira is with Federal Institute of Science and Technology of Paraiba - IFPB, Cabedelo-JP, PB 58100-263 BR (phone: +5583-2101-1721; fax: +5583-2101-1418; e-mail: paulo.ferreira@ifpb.edu.br). and the DOA algorithm are the estimators of major importance in subspace-based methods. The former is used to estimate the number of incoming signals, while DOA estimator is used to estimate the direction of arrival of all incoming signals [1]. An overview of the DOA techniques can be found in [1, 3]. The methods are grouped into beamforming, maximum likelihood, and subspace-based techniques. It shows the performance of algorithms such as capon’s beamforming, SVD, Minimum norm, MUSIC and SPIRIT. Although beamforming techniques are model order insensitive, they may be time-consuming. Moreover, they are limited to resolve closely spaced sources [4], such as S2, S3, S5 and S6 in Fig. 1. Other DOA estimators without estimating the number of sources are inefficient or requires that the number of sensors be at least twice the number of sources [5]. The DOA estimator insensitive to the number of sources (model order) proposed in [5] has improved resolution with the trade- off in the estimation accuracy. Recently, sparse array elements and compressive sensing algorithm have been applied to reduce the time-consuming and complexity of beamforming network [6 - 8]. The number of antenna elements bounds the capacity in model order estimators, which should be equal or greater than the number of sources [9]. The authors in [9] used a pilot signal to increase the capacity of a subarray beamforming- based DOA estimator. Partitioned aperture and subarray techniques have been proposed to increase the array power pattern. The former divided the array into radiating bands [10], while in the latter, the elements are equally spaced [11] and/or overlapped [3, 9]. The constraints are fixed element separation and equally spaced elements, which limit the design of non-uniform reconfigurable elements combination. In [12], the Independent Component Analysis (ICA) was used to identify sources λ0/10 apart, where λ0 is the free space Combining Antenna Array Elements by using ICA Method for Remote Sensing Sources Glauco Fontgalland, Senior Member, IEEE, and Paulo Ixtânio L. Ferreira M Fig. 1. A scenario showing randomly spreaded k sources. It shows sources closely-spaced (S2 and S3), shaded by other (S5 and S6), and far apart radiating on a p-element array.