This paper deals with simple integrator consensus problems. It is well-known that introducing a delay leads in general to a reduction of performances or to instability. Therefore, investigating the effect of time-delays in consensus problems is an important issue. The objective is the design of a improved consensus algorithm for continuous-time multi-agent systems. The novel algorithm proposes to sampled, in an appropriate manner, part of the multi-agent systems information such that the algorithm converges, assuming that at each instant, agent's control laws will also consider the sampled past information of its neighbors. Stability conditions expressed in terms of LMI's and based on algebraic communication matrix structure are provided. The efficiency of the method is tested for different network communication schemes.