The impact of MS velocity on the performance of frequency selective scheduling in IEEE 802.16e Mobile WiMAX. Ashley Mills ∗ , David Lister ∗ , Marina De Vos † and Yusheng Ji ‡ ∗ Vodafone Group Services Limited, Research & Development, Faraday House, The Connection, Newbury, Berkshire, RG14 2FN, ENGLAND. Email: (Ashley.Mills@vodafone.com, David.Lister@vodafone.com) † Department of Computer Science, University of Bath, Bath, BA2 7AY, ENGLAND. Email: mdv@cs.bath.ac.uk § National Institute of Informatics, 2-1-2 Hitotsubashi, Chiyoda-ku, Tokyo, 101-8430, JAPAN. Email: kei@nii.ac.jp Abstract—The OTA performance of Frequency Selective Scheduling (WiMAX Band AMC mode) is compared with that of Frequency Diverse Scheduling (WiMAX PUSC mode) as MS velocity is increased for Mobile WiMAX 802.16e. Frequency Selective Scheduling is shown to outperform Frequency Diverse Scheduling for velocities less than 15km/h and demonstrates upto 50% gain in throughput over the latter. The practical implications of this margin are: that pedestrian MSs in urban deployments may leverage the benefits of fast fading for performance gains without risk. And scheduler implementations can benefit from opportunistic switching between the two schemes given appro- priate differentiating inputs. I. I NTRODUCTION Frequency selective scheduling uses BS (Base Station) estimates of MS (Mobile Station) channel conditions to ap- proximate an optimal allocation of resources to users. It relies on reliable and timely estimates of channel conditions. Its performance is therefore commonly assumed to degrade as MS velocity is increased. The standard proposed alternative at high velocity is to use Frequency diverse scheduling, which spreads user resources over a large bandwidth so as average out the negative effects of frequency selective fading when it is no longer to possible to capitalize on it. We compare the two approaches in 802.16e as MS velocity is increased to de- termine a practical crossover point for differential scheduling strategies. The rest of this paper is organized as follows: Section- II provides some background information, describes the two scheduling approaches, and motivates the case for the study ex- pounded in this paper. Section-III describes prior related work. Section-IV details the technical aspects of the simulations performed, and Section-V presents the results. The results are discussed with relation to practical scheduler decisions in Section-VI and Section-VII concludes. II. MOTIVATION Conceptually, an OFDMA (Orthogonal Frequency Division Multiple Access) frame is a 2D grid of resource units. Each resource unit has a temporal and a subcarrier index, as illustrated in Figure-1. Fig. 1. The resource space provided by OFDMA is a grid indexed by frequency and time. Resource unit a ij has temporal index i and subcarrier index j . Each resource unit corresponds physically to a data symbol. A data symbol is a point on a constellation diagram of some MCS (Modulation and Coding Scheme) 1 . As an example, if QPSK (Quadrature Phase Shift Keying) is being used, then a symbol will take one of the values {00, 01, 10, 11}. The scheduler at the BS is responsible for allocating re- source units within a frame to MSs and for transmitting the frame. The path that the frame takes to the MS will contain ob- stacles which may be moving and which will absorb different wavelengths of EMR (Electromagnetic Radiation), the frame will be subject to other interfering transmissions, Doppler corruption, and reception of the frame will be compounded by multipath reflections. These factors conspire to corrupt each data symbol in the frame individually in a way which depends on the position and motion of the receiving MS relative to the BS, and the exact time at which received frames are transmitted. Thus the 1 Note that 802.16e also uses the term “OFDMA symbol”. This refers to the combination of all symbols in one column and should not be confused. 978-1-4244-5176-0/10/$26.00 ©2010 IEEE This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE CCNC 2010 proceedings