Recent proposals for building scalable quantum computational devices in semi-conductor based spin qubits introduce shared control lines in order to reduce the overhead of qubits controls. In principle, noises from the shared controls could introduce correlated errors to multi-qubit gates, and identifying them will be helpful for achieving higher gate fidelity in those setups. Here, we introduce a method based on the randomized benchmarking protocols that is capable of distinguishing among different correlated noises in a particular two-qubit model motivated by the crossbar architecture.