Phase-alignment conjecture for the three-user semi-symmetric Gaussian interference channel

Consider the three-user semi-symmetric complex Gaussian interference channel

Y1=X1+g1X2+g2X3+Z1,Y_1=X_1+g_1X_2+g_2X_3+Z_1, Y2=X2+g1X3+g2X1+Z2,Y_2=X_2+g_1X_3+g_2X_1+Z_2, Y3=X3+g1X1+g2X2+Z3,Y_3=X_3+g_1X_1+g_2X_2+Z_3,

where Pk=PP_k=P for k=1,2,3k=1,2,3, and write ϕ(g1)\phi(g_1) and ϕ(g2)\phi(g_2) for the phases of g1g_1 and g2g_2. Phase-alignment conjecture. The sum capacity may take its maximum values along the intersection of the following two types of lines:

2ϕ(g1)ϕ(g2)+π=0mod2π,2\phi(g_1)-\phi(g_2)+\pi=0 \mod 2\pi, 2ϕ(g2)ϕ(g1)+π=0mod2π.2\phi(g_2)-\phi(g_1)+\pi=0 \mod 2\pi.

It can be minimized along any of the following two types of lines:

2ϕ(g1)ϕ(g2)=0mod2π,2\phi(g_1)-\phi(g_2)=0 \mod 2\pi, 2ϕ(g2)ϕ(g1)=0mod2π.2\phi(g_2)-\phi(g_1)=0 \mod 2\pi.

The claim is based on numerical observations of the upper bound in several settings; because the observed behavior is not matched by a lower bound, the phase dependence may not be fundamental and the conjecture remains open.

Sources & referencesView supporting material

Primary source

Junyoung Nam, “Capacity Bounds for the K-User Gaussian Interference Channel”, arXiv:1506.03319 (2017).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.