We study three communication primitives in wireless radio networks: Connectivity, One-Receiver, and Gossiping. Radio networks are modeled by undirected graphs of general topology. We consider centralized solutions to the abovementioned problems. In Connectivity and One-Receiver problems, we study the impact of multi-channel assignment to the hardness and approximation of computing of assignments with the minimum number of channels. More precisely, we show that both Connectivity and One-Reciver are Ω(logn)-inapproximable, unless NP ⊂ DTIME(nlog log n). We also give polynomial time algorithms computing multi-channel assignments using at most 3(Δ + ln 2 n) channels for connectivity and at most Δ channels for one-receiver problem, where n is the number of nodes and Δ is the maximum node degree in the graph. Finally, in case of the classical gossiping problem, related to the connectivity problem, we show that it is NP-complete.