You are right. It is caused by physical limit. However, the metamaterial concept open a new window for innovative antenna design including electrically low-profile design with wide bandwidth (low profile is not only issue. the most difficulty is to achieve wideband operation with a low profile.). Very recently, we just finished a design with low-profile but wide bandwidth in terms of impedance matching and gain, a fantastic design indeed.

Always, size vs gain/bandwidth.

Does a metamaterial cloak helps to increase the antenna bandwidth and gain?

Existing cloaking structures all are narrow band and lossy, a game in lab at the moment.

I think this one is the reason : http://en.wikipedia.org/wiki/Chu%E2%80%93Harrington_limit

Thanks for Manish to provide the Chu-limit for small antenna. Is there high hope to overcome that limit by using EBG backed antenna? I did a simulation with a 5mm x 5 mm antenna, without the EBG, it doesn't resonate at 2.6GHz. After adding the EBG, it resonate at 2.6GHz but the gain, bandwidth and efficiency are really poor. Any clue which can help me to overcome the problems in that size?

Physical limits are always there and in order to get close to optimum, you can use the work of R. L. Fante which gives an useful graph (gain, size , frequency and bandwith (Q)).

Fante, R.L., "Maximum possible gain for an arbitrary ideal antenna with specified quality factor," Antennas and Propagation, IEEE Transactions on , vol.40, no.12, pp.1586,1588, Dec 1992

Thanks for the suggestions