A new report has suggested that the various cloaking schemes that have been published so far might actually scatter more EM radiation when integrated over the range of frequencies outside of the cloaking frequency, thus making the object more visible.
According to Andrea Alù and his colleague Francesco Monticone of the University of Texas at Austin, most cloaking techniques used today, including popular ones such as transformation cloaks and plasmonic cloaks, are fundamentally limited by causality and passivity to actually scatter more than the uncloaked object, if you integrate over the entire spectrum, instead of looking at just the wavelength being cloaked. "This means that if you excite the cloak with a pulse, you would actually see it more easily than the uncloaked object it is trying to hide," says Alù. The researchers go on to explain that, apart from the scientific significance of solving the scattering problem, it is equally important for a variety of situations – from warfare to commercial uses – where it is essential that a cloaked object at a given frequency does not become a beacon in a range of the other frequencies.
The authors presented a few suggestions on how to reduce the amount of scattered light. I'm not sure how effect, and more importantly, how realistic those suggestions are based on how these devices are intended to be used.
The paper is published in PRX, which should give you open access to the paper.