Networking scenarios in the future will be complex and will include fixed networks and hybrid Fourth Generation (4G) networks, consisting of both infrastructure-based and infrastructureless, wireless parts. In such scenarios, adaptive provisioning and management of network resources becomes of critical importance. Adaptive mechanisms are desirable since they enable a self-configurable network that is able to adjust itself to varying traffic and channel conditions. The operation of adaptive mechanisms is heavily based on measurements. The aim of this thesis is to investigate how measurement based, adaptive packet scheduling algorithms can be utilized in different networking environments. The first part of this thesis is a proposal for a new delay-based scheduling algorithm, known as Delay-Bounded Hybrid Proportional Delay (DBHPD), for delay adaptive provisioning in DiffServ-based fixed IP networks. This DBHPD algorithm is thoroughly evaluated by ns2-simulations and measurements in a FreeBSD prototype router network. It is shown that DBHPD results in considerably more controllable differentiation than basic static bandwidth sharing algorithms. The prototype router measurements also prove that a DBHPD algorithm can be easily implemented in practice, causing less processing overheads than a well known CBQ algorithm. The second part of this thesis discusses specific scheduling requirements set by hybrid 4G networking scenarios. Firstly, methods for joint scheduling and transmit beamforming in 3.9G or 4G networks are described and quantitatively analyzed using statistical methods. The analysis reveals that the combined gain of channel-adaptive scheduling and transmit beamforming is substantial and that an On-off strategy can achieve the performance of an ideal Max SNR strategy if the feedback threshold is optimized. Finally, a novel cross-layer energy-adaptive scheduling and queue management framework EAED (Energy Aware Early Detection), for preserving delay bounds and minimizing energy consumption in WLAN mesh networks, is proposed and evaluated with simulations. The simulations show that our scheme can save considerable amounts of transmission energy without violating application level QoS requirements when traffic load and distances are reasonable.
展开▼
