- Open Access
Cognitive closed access femtocell application using multi-element antenna
© Islam et al.; licensee Springer. 2015
- Received: 19 May 2014
- Accepted: 2 March 2015
- Published: 28 March 2015
In this paper, a cognitive closed access multi-element antenna-specific femtocell protocol is presented. Femtocell is considered as the preeminent solution for indoor coverage in long-term evaluation (LTE) and LTE-advanced networks. The protocol is verified using a patch antenna for LTE network. In a vast deployment of this mini base station, unwanted handover event is considered as the major obstacle. Access control mechanism with coverage-optimized antenna pattern is one of the most promising solution to resolve this obstacle. A previously proposed microstrip four-element femtocell configuration is used to analyze the performance in user categorizing technique for closed access femtocell network. The categorization is performed using multi-layer feed forward network in neural network. The performance of the technique shows an inherent relationship between completeness of training data and error percentage. Sufficiently low error rates can be achieved by presenting about 15 training samples.
- Multi-element antenna
- Closed access control
- Neural network
Femtocell is a low-power and low-cost data access point that provides high-quality coverage with voice and data services. It communicates with the core network through a wired broadband connection such as cable modem or digital customer line (DSL). The femtocell has a broad prospect in offices and residences as it can handle higher data traffic with less resources allocated than the macrocell. It extends the macrocell coverage at the cell edge regions. However, in dense femtocell network, under existing macrocell coverage, interference and unwanted handover events cause complicacy. In areas with multiple femtocell coverage, signaling congestion increases along with the incessant handover and handoff events. Therefore, disparate solution is required to avoid unwanted handover and interference due to overlapping of coverage. In indoor environment, the performance optimization of the femtocell is mainly dependent on its relative position and allocated resources. The femtocell is usually placed in corners of any residence or apartment, where it is easy to plug in the internet cable (DSL) and the power port. Conventional single-element omnidirectional antenna is less suitable for femtocell coverage optimization as it radiates in all direction at the same intensity. Using multi-element microstrip patch, the antenna mounted on the square/round surface of the femtocell device can be more effective to reduce coverage overlapping.
Access control mechanism has a vast effect on interference level in femtocell network. It can classify the users’ access and the level of service. The mechanism of access-controlled femtocell depends on mainly on the mode selection [1,2]. Three access control modes are used in the femtocell network: open access, closed access, and hybrid access. Selection of each mode enables different mechanisms to serve network users.
Open access allows all the users in the network to connect within the coverage range. Closed access only allows particular users into the femtocell service with less sharing of bandwidth. This particular group is named closed subscriber group (CSG). There is another mode called hybrid access where a limited femtocell service are available for all the users but only the CSG are allowed to use the highest service. In a dense network, open access is a reason for increased number of handover and mobility events. It also shares bandwidth that might decrease the satisfaction level of femtocell owners. It increases the capacity of the network, thus sacrificing network stability. Closed access thus avoids frequent handover and mobility events, but under its superior coverage, it continuously influences the unwanted user to send out handover requests. This induces a new set of cross-tier interference. Hybrid access deals with both problem by adjusting resources according to the number of femtocell owners and subscribers [3,4]. However, for different density of users and femtocell locations, most of the algorithms are outperformed in practical implementation. Then again, owners buy femtocell for their personal benefits. They do not like to share the bandwidth with the random number of user unless they are within their residence.
Studies regarding smart antenna modeling, adaptive beam forming, and DOA estimation are based on the neural network solution [7-14]. Access control mechanism in femtocell is an application of MAC layer. In response to a random user request, it contends the entire signaling process to core network to reject the user up until it is under the femtocell coverage. Restricting the user from the femtocell end with a user categorizing technique will decrease the signaling congestion. In this paper, a neural network-based closed access femtocell network is proposed by using the variations of receiving gains of multi-element femtocell. A previously designed LTE multi-band microstrip antenna  is used for modeling the multi-element femtocell configuration. Results show that the femtocell categorizes the users without any mistake after a certain number of training samples. The rest of the paper is arranged as follows: antenna specification in Section 2, cognitive closed access femtocell in Section 3, simulation and results in Section 4, and conclusions in Section 5.
The power density received by an antenna varies with the incidence angle. While receiving the signal, the antenna responds to an incoming wave from a given direction according to the pattern value in that direction. Each of M number of antenna elements holds different gain patterns in each direction. Therefore, the received power at each element from a particular angle varies due to the prospective antenna gain. Using the variation of received power for the M elements, the femtocell is trained to distinguish the outdoor and indoor users. The conversion of the power from user equipment to the antenna end also depends on the path loss, user equipment class, transmitting antenna gain, multi-path fading, shadowing, noise signals, and receiver port and feed cable losses. However, other terms vary very little within the elements as in practical; femtocells are small in size, with antenna placed very close to each other. The user antennas are considered isotropic that can transmit in all directions evenly. Even if the antenna has directional radiation pattern, it will obtain the M antenna elements location within a small angle from the user end.
Here, H is the conjugate transpose. The femtocell is trained to perform mapping of G:R K → C M from the space of Q(t) to the P(t). The neural network is used to do the inverse mapping F:C M → R K .
where, n = 0,1,2,3…. and T is the sampling period.
The architecture of this paper involves two stages. The initial stage is the training process, that is to learn the traits of indoor and outdoor user using the value of Ē. In the next stage, the network will detect the user based on the previous learning. The network will be trained initially with some sample of indoor and outdoor user’s Ē. After the training, random users are generated. They are considered as isotropic sources. The values of Ē are calculated using Hata path loss model and considering unit shadow-fading and additive white Gaussian noise. When a source radiates towards the femtocell antenna elements, the varying gain of each antenna creates discrepancies in the received power. The neural network exploits these discrepancies to predict the category of the source. The neural network is trained to grant access if the location of the user is inside a certain region . Based on the training experiences, the femtocell distinguishes the indoor and outdoor users. For random values of Ē, the neural network will determine the users’ category by giving an output of ‘1’ or ‘−1’ (sigmoid output neurons). A filtering stage is shown in MN-MUST algorithm that reduces the size of the correlation matrix for cylindrical array configuration for direction of arrival estimation . In this case, the received power is a function of the antenna gain in the impinging direction. For a single user, the more power pattern in each antenna element is considered, the more perfection is achieved in the detection process with less number of training samples.
System parameters for simulation
Randomly generated users (after training)
Femtocell antenna height
User equipment height
UE transmit power (fixed)
Indoor wall loss
Outdoor wall loss
Shadow fading std
White noise power density
This study shows that the reconfigurable coverage area may be achieved by using simple training data that may be generated in real time if necessary. The large single lobe radiation pattern of the antenna with precisely desirable field of view allows the femtocell to operate efficiently. This approach can be applied to any antenna fulfilling the above criteria. This flexible approach to femtocell design is not limited in the area of coverage or total power of transmission. This concept is entirely scalable in terms of the area of coverage. This adaptive technique also serves as an added layer of security over existing wireless systems. Further, it requires very little computation and memory. Thus, it is ideal for implementation in wireless routers, femtocells, or any form of selective wireless broadcasting system.
The femtocell will occupy the major role in providing high indoor coverage in future cellular communication network. Closed access mechanism restricts the unwanted users to connect to primary users’ personal femtocell. A cognitive closed access femtocell operation is presented here using neural network for a multi-element antenna femtocell. The performance of the system in recognizing the environment improves in keeping with the initial training values. The femtocell is trained by a set of trusted primary users. The training will remain valid as long as the femtocell’s position and coverage area are not altered. Thus, any such femtocell will cover any arbitrary region. In the future work, the process will be developed for better femtocell performance with less training illustration.
The research was supported by Ministry of Education (MOE), Malaysia, under the grant scheme no. FRGSTOPDOWN/2014/TK03/UKM/01/1 and ICT fellowship grant for higher education and research in Information and Communication Technology under the Ministry of Post, Telecommunication and Information Technology, Bangladesh.
- G De La Roche, A Valcarce, D López-Pérez, J Zhang, Access control mechanisms for femtocells. IEEE Commun. Mag. 48, 33–39 (2010)View ArticleGoogle Scholar
- AU Ahmed, MT Islam, M Ismail, A Review on femtocell and its diverse interference mitigation techniques in heterogeneous network. Wireless Pers. Commun. 78, 85–105 (2014)View ArticleGoogle Scholar
- X Chu, Y Wu, L Benmesbah, W-K Ling, Resource allocation in hybrid macro/femto networks, in 2010 IEEE Wireless Communications and Networking Conference Workshops (WCNCW), 2010, p. 1–5Google Scholar
- AU Ahmed, MT Islam, M Ismail, M Ghanbarisabagh, Dynamic resource allocation in hybrid access femtocell network. Sci. World J. 2014, 7 (2014)Google Scholar
- H Claussen, F Pivit, Femtocell coverage optimization using switched multi-element antennas, in IEEE International Conference on Communications, ICC’09, 2009, p. 1–6Google Scholar
- H Claussen, F Pivit, LT Ho, Self‐optimization of femtocell coverage to minimize the increase in core network mobility signalling. Bell Labs Tech. J. 14, 155–183 (2009)View ArticleGoogle Scholar
- ZD Zaharis, C Skeberis, TD Xenos, PI Lazaridis, J Cosmas, Design of a novel antenna array beamformer using neural networks trained by modified adaptive dispersion invasive weed optimization based data. IEEE Trans. Broadcast. 59, 455–460 (2013)View ArticleGoogle Scholar
- TG Basha, P Sridevi, MG Prasad, Beam forming in smart antenna with precise direction of arrival estimation using improved MUSIC. Wireless Pers. Commun. 71(2), 1353–1356 (2012)View ArticleGoogle Scholar
- M Agatonović, Z Stanković, N Dončov, L Sit, B Milovanović, T Zwick, Application of artificial neural networks for efficient high-resolution 2D DOA estimation. Radioengineering 21, 1179 (2012)Google Scholar
- D Inserra, AM Tonello, A multiple antenna wireless testbed for the validation of DoA estimation algorithms. AEU Int. J. Electron. Commun. 68, 10–18 (2014)View ArticleGoogle Scholar
- N Fonseca, M Coudyser, J-J Laurin, J-J Brault, On the design of a compact neural network-based DOA estimation system. IEEE Trans. Antennas Propag. 58, 357–366 (2010)View ArticleGoogle Scholar
- J Dudczyk, A Kawalec, Adaptive forming of the beam pattern of microstrip antenna with the use of an artificial neural network. Int. J. Antennas Propag. 2012, Article ID 935073, 13 pages (2012). doi:10.1155/2012/935073Google Scholar
- T Ghouse Basha, A George, B Rajakumar, G Prasad, P Sridevi, A constructive smart antenna beam-forming technique with spatial diversity. IET Microw. Antennas Propag. 6, 773–780 (2012)View ArticleGoogle Scholar
- TG Basha, MG Prasad, P Sridevi, Hybrid technique for beam forming in smart antenna with spatial diversity. Int. J. Wirel. Mob. Comput. 5, 126–136 (2012)View ArticleGoogle Scholar
- AU Ahmed, MT Islam, R Azim, M Ismail, MF Mansor, Microstrip antenna design for femtocell coverage optimization. Int. J. Antennas Propag. 2014, 8 (2014)Google Scholar
- AH El Zooghby, CG Christodoulou, M Georgiopoulos, A neural network-based smart antenna for multiple source tracking. IEEE Trans. Antennas Propag. 48, 768–776 (2000)View ArticleGoogle Scholar
- S Caylar, G Dural, K Leblebicioglu, Neural network method for direction of arrival estimation with uniform cylindrical microstrip patch array. IET Microw. Antennas Propag. 4, 153–161 (2010)View ArticleGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.