This article focuses on the investigation of the Admission Control techniques in order to assure Quality of Service (QoS) in Long Term Evolution (LTE) systems. So as to estimate the users bandwidth requirement, Sounding Reference Signals (SRS) bandwidth are allocated to each user in order to have accurate up-to-date channel state information at the evolved Node B (eNB). Since User Equipments (UEs) are power limited, several SRS are required to scan the full bandwidth for UEs with high path loss. Therefore, in order to own reliable information the eNB must trade-off up-to-date information and accuracy of the measurement. This paper investigates a novel way in which the eNB measures the users radio requirements in order to accomplish QoS, using the information on the SRS already known by the eNB. This measurement technique is compared against two models presented in [1], first a reference AC model in which users are accepted up to a cell bit-rate limit, and a second one proposed by the authors in which the users requirements are estimated using the power control formula and the Shannon capacity. The methodology followed is based on system level simulations. Results show that the SRS based AC is more optimist when estimating users channel conditions and therefore more users are accepted to the system, increasing the overall cell throughput when compared to the power control based algorithm.