Jonathan Segev

Full length article: Advanced coupled PHY and MAC scheduling in IEEE 802.16e WiMAX systems

In this paper we address some issues related to the mutual influence between the PHY layer building blocks (FEC blocks) and the MAC level allocations in IEEE 802.16e /WiMAX systems, in order to increase the overall PHY and MAC combined efficiency. In these systems transmissions are carried in physical Bursts, both on the Uplink and Downlink channels. Bursts are composed of slots, which are grouped into FEC blocks. The number of slots in a Burst determines the length and number of the FEC blocks. The FEC blocks have a direct influence on the probability that bits are received successfully, and thus on the Burst Goodput, which is defined as the ratio between the average number of bits in the Burst that arrive successfully at the receiver, to the Burst length. In this paper we address a new coupled PHY and MAC scheduling methodology by investigating the relationship between the Burst length and its Goodput in different Modulation/Coding schemes, and investigate, given a Burst, the most efficient such scheme. The outcomes of the paper are twofold: first we show that the Goodput of a Burst is almost not dependent on its length. Second, we show that in most cases, the most efficient Modulation/Coding scheme is the one that enables us to transmit the largest number of bits in a Burst. However, there are a few cases where this is not the case. We show these cases in the paper.

Coupled PHY, MAC and repetition scheduling in IEEE 802.16 WiMAX systems

Abstract We address some issues related to the mutual influence between the PHY layer budding blocks (FEC blocks ), the MAC level allocations, and repetition, in IEEE 802.16e/WiMAX systems, in order to increase the overall combined efficiency. We suggest three methods to use repetition: two are based on the increased transmission success probability due to the transmission and decoding of several copies of the same data, and one is based on signal adding and the decoding of one copy with a larger SNR. The last method turned out to be the most efficient one. We show quantitatively that repetition is efficient mainly in low SNRs where otherwise transmissions are not possible, and in particular in negative SNRs. However, there are cases where using repetition is more efficient than using a stronger Modulation/Coding scheme without repetition. Finally, we also show that repetition shall be used with several Modulation/Coding schemes, and not just with QPSK-1/2 as the IEEE 802.16e/WiMAX standard mandates.

Towards an optimal transmission of SDUs in IEEE 802.16e WiMAX systems

Abstract In this paper we address the mutual influence between the PHY layer budding blocks (FEC blocks) and the MAC level allocations in the Uplink and Downlink of IEEE 802.16e systems, and address methods to increase the overall PHY and MAC combined efficiency. We evaluate the efficiency of several schemes for the transmission of Service Data Unit (SDU) packets through Protocol Data Units (PDUs) in physical Bursts. The optimality criterion is the Goodput. When using the Convolutional Turbo Code, Bursts are composed of FEC blocks that have a direct influence on the probability of bits being received successfully, and thus on the Goodput. In the first scheme, every SDU is transmitted as a stand-alone PDU. In the second scheme, the Burst is first divided into PDUs in a way that maximizes the Goodput, assuming the transmission of a bit stream (and not SDUs). Then the SDUs are mapped into the PDUs with possible Fragmentation/Packing. The optimal size of the PDUs is computed by an O ( 1 ) running time algorithm, an important feature because this algorithm is a part of the scheduler in the BS. The results show that when the estimation of the FEC success probability is accurate, SDUs should be mapped into optimal size PDUs in reliable channels. For unreliable channels it is better to transmit each SDU as a separate PDU. When the estimation of the FEC success probability is not accurate, it is always better to transmit SDUs as separate PDUs.