Gauge- and frame-independent decomposition of nucleon spin

In a recent paper, we have shown that the way of gauge-invariant decomposition of the nucleon spin is not necessarily unique, but there still exists a preferable decomposition from the observational viewpoint. What was not complete in this argument is a fully satisfactory answer to the following questions. Does the proposed gauge-invariant decomposition, especially the decomposition of the gluon total angular momentum into its spin and orbital parts, correspond to observables which can be extracted from high-energy deep-inelastic-scattering measurements ? Is this decomposition not only gauge-invariant but also Lorentz frame-independent, so that it is legitimately thought to reflect an intrinsic property of the nucleon ? We show that we can answer both of these questions affirmatively, by making full use of a gauge-invariant decomposition of covariant angular momentum tensor of QCD in an arbitrary Lorentz frame.