In this dissertation, we discuss an efficient multiple access control protocol based on packet transmission technology and efficient handoff schemes with priority for handoff attempts in the next generation mobile communications networks.
First, we propose the fast packet reservation multiple access (PRMA) scheme, which is a mobile-to-base transmission protocol. The original PRMA, which is a reservation multiple access protocol performing statistical multiplexing at the talkspurt level, has its inherent disadvantage that its performance deteriorates abruptly under high offered load. Mitrou's PRMA proposed to overcome this shortcoming gives the voice traffic longer delay than the original PRMA, because the MS, which succeeds in contending on the first available reservation minislot, is not permitted to transmit its information packets before the start of the next frame. The fast PRMA proposed by us gets over these two disadvantages efficiently. We analyze the performance of the fast PRMA in two cases: the voice-only fast PRMA and the voice/data integrated fast PRMA. For the performance evaluation of the voice-only fast PRMA, we use a Markov analysis. From the numerical results, it is shown that the fast PRMA provides the expanded capacity and the decreased average delay for the voice traffic. For the voice/data integrated transmission schemes, most of those multiple access protocols have a defect that accommodating data traffic deteriorates the QoS of the voice traffic. Besides, most of the works on the voice/data integrated transmission assume that each data terminal has a storage packet buffer, the size of which is one. In the voice/data integrated fast PRMA, each data terminal is assumed to have an infinite storage packet buffer and the existence of the data traffic sources does not deteriorate the QoS of the voice traffic sources. We use a Markov analysis for the voice subsystem and an equilibrium point analysis (EPA) technique for the data subsystem in order to create an analytic formula. As performance measures, the voice packet dropping probability, the average voice and data delay, and the total throughput are drived.
Lastly, we propose two queueing priority channel access assignment schemes in a hierarchically overlaid system with microcells and macrocells: the partial buffer sharing (PBS) scheme and the dual queue length threshold (DQLT) scheme. In the two-tier system, the microcell tier is dedicated to the low-mobility MSs and the macrocell tier is dedicated to both the high-mobility MSs and the low-mobility MSs overflowed from microcells. We calculate the handoff attempt rates both in microcells and in macrocells. The overflow traffic from the subordinate microcells to the macrocells is modeled as an (N+1)-state Markov Modulated Poisson process (MMPP). Effects of the size of a macrocell and of hierarchical cell structure on system performance are discussed when the proposed PBS and DQLT schemes are used as channel assignment schemes with priority at the macrocell tier. The performances of the proposed schemes are also compared with those of the no priority scheme (NPS) and the first-come-first-service (FCFS) queueing schemes. From the numerical results, it is shown that these proposed schemes accommodate low- and high-mobility MSs with the lower forced termination probability. It is also shown that the PBS scheme has the best performance among the four schemes.
