ABSTRACT The light-cone effect causes the mean as well as the statistical properties of the redshifted 21-cm signal ${T_{\rm b}}(\hat{\boldsymbol {n}}, ν)$ to change with frequency ν (or cosmic time). Consequently, the statistical homogeneity (ergodicity) of the signal along the line-of-sight (LoS) direction is broken. This is a severe problem particularly during the Epoch of Reionization (EoR) when the mean neutral hydrogen fraction ($\bar{x}_{\rm {H\,{\small I}}}$) changes rapidly as the Universe evolves. This will also pose complications for large bandwidth observations. These effects imply that the 3D power spectrum P(k) fails to quantify the entire second-order statistics of the signal as it assumes the signal to be ergodic and periodic along the LoS. As a proper alternative to P(k), we use the multifrequency angular power spectrum (MAPS) ${\mathcal {C}}_{\ell }(ν _1,ν _2)$, which does not assume the signal to be ergodic and periodic along the LoS. Here, we study the prospects for measuring the EoR 21-cm MAPS using future observations with the upcoming SKA-Low. Ignoring any contribution from the foregrounds, we find that the EoR 21-cm MAPS can be measured at a confidence level ≥5σ at angular scales ℓ ∼ 1300 for total observation time tobs ≥ 128 h across ∼44 MHz observational bandwidth. We also quantitatively address the effects of foregrounds on MAPS detectability forecast by avoiding signal contained within the foreground wedge in $({\boldsymbol {k}}_⊥ , k_∥)$ plane. These results are very relevant for the upcoming large bandwidth EoR experiments as previous predictions were all restricted to individually analysing the signal over small frequency (or equivalent redshift) intervals.