ISM antenna on two sides of device: how to design?

I am designing a device that operates in the ISM band (2.4-2.48) for which omnidirectionality and size are important. The device has a cylindrical shape, with a stack of battery, PCB and speaker (in that order). This makes the placement of the antenna tricky: in the center on the PCB the signal will be blocked by the speaker and battery, and letting the PCB extend beyond the speaker and battery (in diameter) is not possible. So, the solution I have so far: the PCB is a flex PCB consisting of three circles. The center one is the normal PCB part, the two other circles contain the antenna. One of those folds on top of the battery-PCB-speaker stack, and one to the bottom. This way the signal isn't blocked and should be rather omnidirectional. A sketch of this PCB is attached.

But, now the design of the antenna itself. What I have now: on the PCB the RF signal is created. This signal is put on a PCB trace with a length of about 1 lamba to the top circle and 1 lambda to the bottom circle. The circles all have a diameter of 20mm, so on the top and bottom circle these traces meander a bit. I simulated this contraption in CST, but the best S parameter I get is -6dB. the omnidirectionality is pretty good though. I expect that this can be improved, but I have very little experience with antenna design, and because this antenna consists of two parts (one to the top and one to the bottom with the feed in the middle), I'm not sure how to use a different of the shelf solution. Who has a better idea?

I would use a loop antenna around the periphery of the board. If that does not fit, try just an antenna as far to one side of the board as possible and push everything else to the other side.

A loop antenna would be possible, but the diameter of the PCB is too small to make it work at 2.4 GHz (D=20), and aside from the battery and speaker on the top and bottom of the PCB: there is an aluminum ring around the PCB. If necessary this can be changed if it really gives a good antenna, but I prefer to keep it.

in the center on the PCB the signal will be blocked by the speaker and battery

Click to expand...

Possible but seldom true. Speaker is probably not all metal, and neither battery or speaker are not screening RF like a paper sheet in front of a light-bulb, creating sharp shadows.
If there is any blocking effect at all is depending on such thing as battery size relative wavelength. Battery can as well be a director as reflector.

An antenna must have both a certain height above RF ground and and reasonable length to perform well. There are no shortcuts from this.

Design everything of a electronic circuit including cute decorative metal rings, and then starts thinking, where to place the antenna? Such design level make me mad.
It is as designing a hand-torch and forgetting to leave space for the bulb/LED, but it have implemented cute decorations with high design status. If someone comes to my lab, asking me to do a such antenna-job, I will kick them out. Hard. Main reason is that neither me or my customer will ever be satisfied with the antenna work that is possible for me to do. It is poor design from start and it cant be repaired with a miracle antenna.

Most of the electronic parts can be placed in a lot of alternative places and be given different shapes, including battery, PCB and speaker. That is why it is better to decide place and size for the other parts that not can be reshaped without suffer in performance before thinking about decorations.

To get an idea about basic limits for this antenna design, if antenna is placed in the both/either ends, what height above battery will antenna have, guess battery is some form of RF ground?
What is the available antenna height at speaker side?
Is final product something that could be hold with a hand or similar? If so will it heavy reduce antenna performance if the antenna is placed to close where the hand will hold the unit. Can the antenna be given a position that gives it RF height without being routed close to these areas, such as more central in the tube? Diameter is 20 mm, but if we include that antenna not should totally die due to hand effect, is available space for antenna still a circle with 20 mm diameter?

If hand effect not is relevant in this case can antenna be a monopole at the edge of PCB. Leave as much space as possible, in all directions seen from antenna, free from grounding objects and you have an reasonable antenna. That is most simple and most common solution for small units in this frequency range. Probably not peak performance but good enough.

Speaker coil can in some cases also act as antenna at these frequencies. Dual functions for the coil and spacesaving. I have done one or two such designs just because some less smart designer did forgot to reserve reasonable space for the antenna from start. It do seldom result in a good antenna as the coil wire is resistive lossy at these frequencies, but better then nothing.
It is about the same thing, using battery as part of an antenna. It is rather easy split battery wires in one DC circuit and one RF circuit but it is lossy and unstable as wires moves around.

In most cases is final product more compact and better performing if it is correct designed and dimensioned from beginning, with ALL components included. From RF view, start the design with the most important component, the component that set the limits whatever RX/TX chip that then is used or how effective the battery is.

In this case, why not reduce battery size with 50% (3dB) if needed to leave space for a better antenna. As example, from 2 to 7 mm available antenna height can the difference be more then 6 dB in antenna efficiency, if that space can be given by reduce battery size can it be a good deal. It can increases total battery hours, as less power is needed in TX circuit while having better coverage. Also positive if reduced weight is of interest.