Design and Configuration Techniques of a Low-Profile Reconfigurable Antenna for a Cognitive Radio System Joseph A. Zammit, Adrian Muscat Department of Communications and Computer Engineering University of Malta Email: jzam0022@um.edu.mt,adrian.muscat@um.edu.mt Abstract Cognitive mobile radios in the longer term will require broadband small antennas spanning the UHF and lower microwave mobile radio bands, typically 400MHz to 3Ghz. For these bands to be covered the antenna requires technology that allows hard switching and soft tuning mechanisms both in the radiating structure as well as in the feeding structure to achieve the desired bandwidth. These antenna controller is expected to learn in an efficient way how to tune the antenna that is under its control with the influence of the physical and operative radio environment. One way of implementing a tuning algorithm in a fully reconfigurable antenna is to make use of a heuristic algorithm. However this technique will result in a large search space. In this paper a fully reconfigurable antenna consisting of a pixel patch antenna is presented and several sets of simulations are presented each showing how the antenna can be tuned to a desired frequency.in this first order antenna model several techniques are suggested to prune the resulting search space. The methods are derived from three distinct sets of simulations that take into consideration the effects of sub-patch hard switching, varactor diode loading and shorting post loading. The results show that such an antenna is feasible. Index Terms antenna, cognitive radio, re-configurable, antenna switch 1. Introduction As more wireless services are becoming more common, the available radio spectrum is decreasing even as higher frequency bands are being opened for use. Another issue is that different countries have their own spectrum allocations and spectrum management techniques which restrict the user s mobility. Cognitive Radio technology tries to overcome these issues by adding a level of computational intelligence to the terminal to provide a seamless service and dynamic service. 1 shows the hardware architecture of a modern Software Defined Radio (SDR). Figure 1. Hardware Architecture of Software Defined Radio [1] Many advances have been made in the field of microelectronics to optimize the RF front end of the SDR. However modern commercial radio systems are limited to a few bands such as tri-band mobile phones or the SpeakEasy [1] system. A cognitive system needs to have access to the full radio spectrum to be able to operate in an optimal manner. Currently the limiting factor in the access to the radio spectrum has been the antenna. Antennas have been traditionally designed to operate on a single frequency or to be switched between a few bands. Reconfigurable antennas aim to fill this gap by being able to switch between various frequencies. Various attempts are described in [2], [3], [4] with a degree of success. However a new class of reconfigurable antennas is needed which can tune from the UHF to the upper microwave band (0.4GHz to 3GHz). Moreover antenna control algorithms for rapid switching and tuning need to be developed. However
before this can be achieved, the various antenna parameters need to be adequately represented in order for the best solution to be found. This paper describes attempts to design highly reconfigurable antennas by the means of ideal switches (hard tuning) and varactor diodes (soft tuning). A fully reconfigurable pixel patch patch antenna is presented and techniques are proposed to reduce the search space generated by such a reconfigurable antenna. but retained a similar pattern. Due to its small size the antenna had a low overall gain. 2. Reconfigurable antenna techniques There are two main methods where the resonant frequency of the antenna may be changed. The first is using switches, usually pin diodes [3] or MEMS switches [5]. This is known as hard switching and produces an antenna which can switch through various bands while retaining good gain and efficiency. Soft tuning techniques utilize reversed biased varactor diodes in order to continuously change the resonant frequency [?]. Such a method is easy to control but produces antennas with a very low gain and efficiency. Many authors have worked on reconfigurable antennas. The slot antenna built by Behdad [2] is a varactor tuned slot antenna which can resonate from 1.5GHz to 3GHz. Nikolaou [3]using an annular slot antenna could switch between 5.8GHz and 6.4GHz. Also the polar pattern could be reconfigured as well. Holland [4] built a tunable coplanar antenna that had a 500MHz bandwidth. Patch antennas could also be frequency reconfigured such as in the case of Chung et al [6], the patch antenna had a U shaped notch which could be shorted using a pin diode. The antenna could switch between 2.4GHz and 2.62GHz. The antenna was built but the pin diodes were substituted by metal tapes, hence their effect on the antenna was not included in the experiment. However these antennas though they had a good performance only had a very limited tuning range. Cetiner at al [7] describe a pixel-patch reconfigurable antenna which can change its frequency and polarization. The author has investigated both soft tuned and hard tuned antennas and a highly reconfigurable antenna consisting of both soft and hard tuned elements [8], [9], [10]. All the antennas produced good results which are under further investigation. In [8] a small antenna measuring 22x20mm was developed. The antenna could be tuned from 0.55GHz to 1.5GHz. Such low resonant frequencies were obtained by the use of shorting posts in line with varactor diodes as shown in Fig. 2. The radiation pattern of the antenna changed with decreasing resonant frequency Figure 2. Varactor Tuned shorted post antenna A different approach was taken in [9] where the antenna was reconfigured by switching parasitic coupling patches. A total of 13 switch combinations covered a bandwidth from 1.36GHz to 1.92GHz. However such a system left some frequency bands which were not covered. The advantage of using switched coupling patches was that the antenna exhibited a higher efficiency than the soft tuned antenna. Figure 3 shows the frequency response of the antenna structure. Figure 3. Frequency response of the parasitic switched patch antenna Both techniques using soft tuning and hard switching were demonstrated in [10]. The proposed structure consisted of switchable shorting posts and varactor diodes. The structure of the antenna is shown in Fig 4. The antenna resonated from 0.55GHz to 0.7GHz
and from 1.3GHz to 2GHz. However there was a large variation in the gain and efficiency over the whole band. Figure 5. Fully Reconfigurable antenna Number of vertical switches = P w (P l 1) (2) Total number of switches in matrix = (P w 1)P l + P w (P l 1) Figure 4. Reconfigurable antenna using hard and soft tuning techniques = P w P l P l + P w P l P w = 2P w P l (P l + P w ) (3) 3. Antenna Design In order to have a fully reconfigurable antenna, the antenna needs to be able to change its shape to be resonate at the required frequency. In order to achieve this a series of square patches were fully interconnected together using ideal switches. Figure 5 shows the whole concept. The patches were selected to be 4mm x 4mm each and an array consisting of 6 x 4 sub patches was constructed where each sub-patch was fully connected to its neighbours. The following is the derivation of the number of switches needed for full interconnection. Switches along the width = P w Switches along the length = P l Number of horizontal switches = (P w 1)P l (1) If P l = P w than equation 3 becomes 2P 2 w 2P w (4) It can be noted that equation 4 has a complexity of O(n 2 ). Thus a large number of switches will be needed in order to increase the reconfigurability of the antenna. In this case the antenna has a total of 38 switches. 4. Experiments and results Two types of configurations were run on the antenna. In the first case the switches were reconfigured in a way to emulate several classic antennas, in the second case a dipole shape was chosen and several parasitic elements were added to it in order to vary the resonant frequency. The configured shapes are shown in Fig. 6. Simulations were performed using CST microwave studio [11]. The antenna was simulated on a Rogers 4003 substrate having an ε R of 3.38mm and a height
Figure 6. Reconfigurable antenna regular shapes Table 1. Resonant frequency for various shapes Antenna Shape Rectangular H shape Small Octagon Octagon Patch with slots Dipole Meander line Patch Short Dipole Square Figure 7. Resonant frequency for regular shapes of 1.524mm. A small ground plane 80x40mm was used to simulate the size of a mobile terminal. For all simulations the antenna was fed from at the centre of the patch by a probe feed. Figure 7 and 1 show the results that were obtained when the antenna was reconfigured into various regular shapes. It can be seen that a high gain and antenna efficiency were obtained over a large bandwidth stretching from 2.56GHz to 3.92GHz. Next the dipole antenna was selected and the effects of switching on various parasitic patches next Resonant Frequency (GHz) 2.56 1.96 4.17 2.77 4.3 2.59 3.57 3.92 3.82 S11 Gain Efficiency (%) -3.4-11.6-32.5-4.9-11.62-11.47-27.81-17.73-13.87 4.5 3.3 6.04 4.05 4.1 4.1 7.26 6.1 6.8 39 54 33 41 56 57 84 75 81 to the dipole were observed. The position and number of parasitic patches were varied. Figure 8 shows the combinations of the dipole which were selected. The results are shown in table 2 and the return loss in figure 9. It can be noted that by switching on and off parasitic patches the resonant frequency can be changed. 5. Discussion of results From the above results it can be conclude that a reconfigurable antenna made up of an array of subpatches can be made to change its resonant frequency by either reconfiguring the whole shape of the antenna or by adding parasitic patches to a standard shape. This
Figure 8. Reconfigurable antenna regular shapes Table 2. Resonant frequency for parasitic additions to the dipole Dipole Figure 9. Return loss of the added parasitic patches to the dipole process will happen dynamically in the terminal while it is searching for a unused part of the radio spectrum. However the large number of switch combinations make it lengthy operation to search in real time for the optimal solution. As shown in 4 the search space increases in a non-linear fashion and techniques to prune the resulting search space need to be used. The simplest method would be to exhaustively search all switch combinations and produce a look up table which can be referenced to select the required frequency. However to make an exhaustive search would take a long time to complete due to the large number of simulations required. Also such an antenna 1 2 3 4 5 6 7 8 9 Resonant Frequency (GHz) 5.87 2.27 2.09 5.65 2.44 2.2 2.3 2.25 2.26 S11 Gain Efficiency (%) -11.5-4.9-12.9-9.1-8.6-12.0-1.5-11.3-10.4 5 3.6 3.8 6.2 3 3.7 1.8 2.42 2.4 78 40 56 78 45 57 14 43 42 will not be able to adapt to the radio environment dynamically. Other techniques such as the use of genetic algorithms [?] and simulated annealing [?] may be used to find a particular resonant frequency. A weighted fitness function consisting of the the following parameters maybe used. For a real time system parameters 3,4,5 made be more difficult to measure. 1) Resonant frequency 2) Return loss 3) -10dB bandwidth 4) Gain 5) Efficiency The switches could be represented by bits in the genome so for example the antenna presented in this
paper could be represented as a genome with a length of 38 bits. The optimal way to find the best combinations in the search space is to use evolutionary algorithms and case based rules to prune the search space. A method to reduce the search space is to ensure that the when a switch is put in an on position a path exists between the switch and the antenna probe. Another method is for the search algorithm to estimate the length and width of the patch to produce the desired resonant frequency and then add/subtract parasitic patches to produce the desired result. Also results that have produced an optimized antenna could be stored to be used in future cases. The reconfigurable patch suggested in this paper is made up is square sub-patches and is a regular shape. For specialized applications irregular sub-patches may be used to produced the desired result. It can be also suggested that and offline evolutionary algorithm could be used to develop the antenna structure, which is turn can be used by other evolutionary algorithms during operation to configure the antenna. 6. Conclusion This paper has discussed the current literature in reconfigurable antennas. It has shown a fully reconfigurable pixel-patch antenna made up of an array of square sub-patches and how they can be re-configured to resonate at different frequencies. It has also demonstrated how parasitic patches can be added so that the resonant frequency changes slightly. The results show that this can be achieved though the matching is very poor. This could be solved by adding a tunable impedance matching network as suggested by Mingo et al. [12]. The resultant reconfigurable antenna consists of a large number of switches which means a large search space in order to find the optimal solution. Several techniques are suggested on how to prune and search this space. Acknowledgements The authors wish to thank the University of Malta for providing funds for this research project. References [1] B. Fette, Cognitive Radio Technology. Newnes, 2006, ch. 1. [2] N. Behdad and K. Sarabandi, Dual-band reconfigurable antenna with a very wide tunability range. [3] S. Nikolaou, C. Bairavasubramanian, R.Lugo, I. Carrasquillo, D. Thompson, G. Ponchak, J. Papapolymerou, and M. Tentzeris, Pattern and frequency reconfigurable annular slot antenna using pin diodes, IEEE TRANSACTIONS ON ANTENNAS AND PROP- AGATION, vol. 54, no. 2, pp. 439 448, 2006. [4] B. R. Holland, R. Ramadoss, S. Pandey, and P. Agrawal, Tunable coplanar patch antenna using varactor, Electronics Letters, vol. 42, no. 6, 16th March 2006. [5] P. Panaiat, C. Luxey, G. Jacquemodt, R. Starajt, G. Kossiavas, L. Dussopt, F. Vacherand, and C. Billard, Mems-based reconfigurable antennas. IEEE International Symposium onindustrial Electronics, May 2004. [6] Y. N. Chung, T. Yun, and C. J., Reconfigurable microstrip patch antenna with switchable polarization, IEEE Microwave and Wireless Components Letters, vol. 12, Mar. 2002. [7] B. Cetiner, H. Jafarkhani, J.-Y. Qian, A. Hui Jae Yoo; Grau, and F. De Flaviis, Multifunctional reconfigurable mems integrated antennas for adaptive mimo systems, IEEE Communications Magazine, vol. 42, no. 12, pp. 62 70, December 2004. [8] J. A. Zammit and A. Muscat, A small tunable antenna using multiple shorting posts and varactor diodes. Malta: ISCCSP 2008, Mar. 2008. [9], Tunable microstrip antenna using switchable patches. University of Loughborough: LAPC 2008, Mar. 2008. [10] A. Muscat and J. A. Zammit, Reconfigurable antenna structure for a wideband cognitive radio. The Institution of Engineering and Technology, Sep. 2008. [11] Cst microwave studio http://www.cst.de. [12] J. Mingo, A. Valdovinos, A. Crespo, D. Navarro, and P. Garcia, An rf electronically controlled impedance tuning network design and its application to antenna input impedance automatic matching system, IEEE Transactions on Microwave Theory and Techniques, vol. 52, pp. 489 497, February 2004.