Modern real-time information dissemination services generate data dynamically and process data requests on demand. Such data or requests are usually valid or useful subject to temporal conditions. This paper devises an efficient on-demand time-critical data broadcast scheme by considering request urgency, service productivity and access fairness, to maintain quality of service and performance. The experimental studies show the proposed scheduling approach is able to achieve low request deadline miss rate and mean access time simultaneously. Its performance is superior to classical EDF and MRF and further better than SIN-k and RxW methods by simply tuning parameters.