This paper proposes a scheme for hard handover (HO) between base stations (BSs) that combines bicast with forwarding; it realizes packet-lossless HO as well as low HO control delay. The proposed scheme observes the status of the current channel condition and initiates bicasting, the simultaneous transfer of IP packets from the access router to both the old BS and the new BS, when the probability of HO becomes high; this reduces the control delay imposed by hard HO. When HO becomes unavoidable, only those IP packets remaining in the old BS buffer that are not shared with the new BS are forwarded to the MS; this prevents the loss of IP packets. Computer simulations show that the proposed scheme reduces the HO control delay at 95% cumulative distribution function (CDF) by approximately 1700 (300) msec, 340 (320) msec, and 170 (330) msec compared to the conventional forwarding scheme (the conventional bicast scheme) when the number of users is 80 and the maximum Doppler frequency (fdmax) is 5.55 Hz and data rate (D) on the wired propagation channel is 10, 50, and 100 Mbps, respectively. The results confirm the superiority of the proposed scheme as an IP packet loss prevention scheme for hard HO.

This paper proposes cell selection (CS) based on shadowing variation for the forward-link Orthogonal Frequency and Code Division Multiplexing (OFCDM) packet wireless access. We clarify its effects using a broadband propagation channel model in a comparison with fast cell selection (FCS), which tracks the instantaneous fading variation, and with the conventional slow CS, which tracks only the distance-dependent path loss, based on radio link level simulations that take into account time-varying instantaneous fading and shadowing variations. The simulation results show that the achievable throughput with FCS improves slightly in a broadband channel with an increasing number of paths when the average path-loss difference between two cells is greater than 2 dB. Nevertheless, we show that the optimum CS interval becomes approximately 100 msec, because the interval can track the time-varying shadowing variation considering low-to-high mobility up to the maximum Doppler frequency of 200 Hz. Consequently, we show that the throughput by employing the CS based on shadowing variation with the selection interval of 100 msec is increased by approximately 5 and 15% compared to that using the conventional slow CS with the selection interval of 1 sec, for the maximum Doppler frequency of 20 and 200 Hz, respectively.