In wireless networks, there are two main scheduling problems: uplink (mobile station to base station) and downlink (base station to mobile station). During the downlink scheduling, scheduler at the base station (BS) has access to queue information of mobile stations (MS). On the other hand, for uplink scheduling, BS only has the partial information of the MS since distributing the detailed queue information from all MSs to BS creates significant overhead. In this thesis, we propose a novel uplink scheduling algorithm for delay sensitive traffic in broadband wireless networks. In this proposed algorithm, we extend the bandwidth request/grant mechanism defined in IEEE 802.16 standard and send two bandwidth requests instead of one: one greedy and the other conservative requests. MSs dynamically update these bandwidth requests based on their queue length and bandwidth assignment in previous frames. The scheduler at the BS tries to allocate these bandwidth requests such that the system achieves a high goodput (defined as the rate of error-free packets delivered within a maximum allowed delay threshold) and bandwidth is allocated in a fair manner, both in short term and in steady state. The proposed scheduling algorithm can utilize the network resources higher than 95% of the downlink scheduling algorithms that use the complete queue state information at the MS. Using just partial queue state information, the proposed scheduling algorithm can achieve more than 95% of the total goodput achieved by downlink scheduling algorithms utilizing whole state information. The proposed algorithm also outperforms several downlink scheduling algorithm in terms of short-term fairness