This paper examines algorithms to minimize interference in infrastructure wireless mesh networks. A mathematical optimization problem for coloring the active wireless edges in a time-division scheduling frame is formulated. The optimization eliminates primary conflicts and minimizes secondary conflicts. The active wireless edges are specified in an integer zeroone 'edge-specification' matrix. An 'edge-interference' matrix is formulated, where each element represents the interference power if two edges share a color. The objective of the optimization problem is to partition the integer edge-specification matrix into a sum of C integer zero-one matrices which specify the active edges assigned to each of the C colors, such that the secondary interference is minimized. The optimal solution requires a constrained partitioning of an integer matrix, which is a combinatorial problem. A polynomial time approximation algorithm called Least-Noise coloring is presented. Simulations of an essentiallysaturated hexagonal mesh network supporting backhaul traffic flows are reported. It is confirmed that interference and edge transmission powers can be minimized, and that the mesh network can be configured to achieve near-perfect Quality of Service guarantees with essentially 100% throughput.