Help with MSK modulation

I have 5 bursty bit streams which are to be modulated by MSK at different carrier frequencies.
The bit stream is zero ( if no signal is there) , or 1 or -1 ( 1-> logic 1 and -1 ->logic 0). each symbol ( 0 , -1 or 1 ) is sampled 8 times with the sampling interval = 4e-7 seconds.

How can I modulate these streams in MATLAB ?
After modulating them at different carrier frequencies (890,895,900,905,910,915 MHz) I have to sum them up ... or multiplex them .. to simulate the combined air channel traffic to be filtered into different bands later on using bandpass filters.

I have tried Simulink - but the MSK or the GMSK blocks do not ask for a carrier frequency specification.
The modem.mskmod function just converts my bit stream (consisting of only 1 and 0 ) into complex values ( which form the new signal constellation) which i don't know how to use to modulate using a carrier wave of the specified frequencies.
Please help !!

PS : And one more doubt ==> to simulate GMSK .. do i need to first filter the NRZ bit sequence using gaussian filter and then MSK modulate it ... OR .. vice-versa ?

MSK is pretty straightforward using matlab: MSK. You can also develop the code since it is a simple multiplication of sines and cosines to the data to create continuity. The details of which, can be found in most digital comm books e.g., proakis, sklar, mk simon, haykin.

All you need is to have a look at the communication toolbox of matlab, where you can sum these signal. The bandpass filtering you can do in the frequency domain by using the filter function or the convolution function for more in depth control.

When using simulink to modulate to the carrier frequencies you need to e.g. use FDM.

GMSK => shape the data with a gaussian filter then use the MSK to modulate.

Have fun!

oh okay , so mskmod would create the complex envelope , which i can then multiply with a carrier wave ( e.g. cos (2*pi*890e6*t)) to have my msk modulated stream. and i wont have to do any odd and even sequence separationa dn then the separate multiplication. Am i Right ?

secondly , if i wanna go for GMSK .. the data i get after the gaussian filter is not binary , and hence would not be fit to be modulated using mskmod ( it takes only binary input data). How should I go about Msk modulating then?

Ps : Thank you for your help.

i) your idea for carrier formulation seems mathematically right.
ii) GMSK techniques are available widely GMSK.
GMSK implies your using a gaussian pulse shaping instead of the half sinusoid, so there really should be no issue.