Soft handover is one of the most important merits of In recent years, wireless mobile communication systems have experienced tremendous growth. To utilize the radio spectrum efficiently, the cellular architecture is used in wireless mobile networks. In such networks, the cell coverage and capacity of a network are planned in the pre-deployment stage ac-cording to pre-defined traffic patterns. In practice, however, traffic patterns are changing with time due to random user mobility and versatile service activity. Therefore, the planned cellular mobile networks may not utilize radio resources optimally under the varying traffic patterns. In next-generation code division multiple access (CDMA) cellular networks, this problem becomes more severe especially for the downlink (from base station to mobile sta-tion) capacity-limited scenario because of the necessity for abundance downlink multimedia traffic and the interdependence of coverage and capacity in CDMA systems [1], [4], [29], [53].
In response to the variation of traffic patterns, tradeoffs between coverage and capacity should be considered carefully in CDMA cellular systems [1], [29], [30], [54]. For example, to guarantee the coverage of a cell, more power is used for mobile users near cell bound-aries under power control. However, this will generate higher inter-cell interference to other cells, which reduces the system capacity significantly. Moreover, power control may not be effective if a large traffic variation occurs [1], [29], [30], [54], [55]. It is found in [54] that uniform network layout with equal-sized cells is optimal for uniform distributed users, and the capacity degrades significantly if traffic loads are not balanced over all cells. Therefore, to utilize radio resources efficiently, it is crucial for next-generation CDMA cellular networks to be aware of system situations and configures cell coverage and capacity dynamically to balance traffic loads over all cells [29], [30].
Several schemes have recently been proposed for dynamic cell configuration in cellular networks [31]−[37]. In [31], the optimization of pilot power and the planning procedures of downlink capacity and cell coverage were proposed. In [32], authors used analytical methods to study the competitive characteristics of network coverage and capacity in a simple network.
Only one class of service was considered in [31] and [32], and it may be difficult to extend
these schemes to a network with multi-classes of services. There are also some heuristic-rule-based techniques in the literature for dynamic pilot control to balance downlink traffic load while assuring service coverage [33]−[35]. However, these schemes may cause some
“coverage failure regions” between cells where all the received pilot signals are too weak to serve a mobile station [36], [37]. Moreover, the common shortcomings of the previous work [31]−[37] are that only pilot power is adjusted and other radio resource management schemes are not taken into account in the time-varying environment.
In fact, pilot power allocation and other radio resource management schemes, such as soft handoff power and maximum link power allocations as well as call admission control mechanisms, are highly coupled in CDMA systems. For example, it was showed that sig-nal quality degradation can be prevented by configuring cell areas adaptively and setting transmission power levels appropriately [4]. Also, authors in [38] and [39] showed that soft handoff has significant impacts on the system capacity and cell coverage. Moreover, we have presented an effective link proportional power allocation (LPPA) for soft handoffs in chapter 2, 3 and [8], which can enhance system capacity in mixed-size cellular systems compared to the conventional site-selection diversity transmission (SSDT) scheme [14].
In this chapter, we show that dynamically adjusting pilot power alone while not changing other radio resource management algorithms accordingly can result in performance degra-dation. We then propose a novel reinforcement-learning approach to solve the dynamic cell configuration problem in multimedia mobile CDMA networks. The novelties of the proposed scheme are as follows.
1. The proposed scheme takes into account pilot, soft handoff, and maximum link power allocations as well as call admission control mechanisms, enabling it to dynamically adapt to changes in traffic situations and improve the system performance [56].
2. It can efficiently tackle problems with large state spaces and action sets by applying reinforcement-learning algorithms [57]. Since there will be several service classes in future CDMA networks, the state spaces and action sets are very large in the dynamic cell configuration problem. It was shown that reinforcement-learning algorithms make
it feasible for the first time to solve optimization problems for large-scale realistic networks, which were previously deemed intractable [58].
3. The proposed scheme can be implemented in a distributed manner in each base station, which monitors the variation of its power load that can implicitly reveal the load information about all other cells in the whole network. Therefore, the coverage and capacity can be coordinated between cells accordingly. The system can thus be fully self-organized for dynamic cell configuration.
4. The distributed algorithm can avoid overloaded signaling between base stations and ra-dio network controllers, which is necessary for the centralized algorithm, in the systems with dynamic cell configuration. Besides, the modelling of the centralized algorithm needs volumes of system states, which results in high computation complexity to obtain an optimal solution that may be outdated. Therefore, the efficiency of the distributed algorithm makes the system be adaptive to the situations with greatly traffic variation.
5. It does not require a priori knowledge of the state transition probabilities associated with the cellular networks, which are very difficult to estimate in practice due to the different propagation environment, diverse multimedia services, and random user mobility. Therefore, the assumptions behind the underlying system model can be made more realistic than those in previous schemes.
We compare our scheme with the fixed scheme and the scheme in which only pilot power is adjusted dynamically but other radio resource management algorithms are not changed accordingly. Extensive simulation results show that the proposed dynamic scheme outper-forms the others by increasing the total throughput, decreasing the frame error probabilities, blocking probability, and handoff forced termination probability with the price of increasing the size of active set slightly.
The rest parts of this chapter are arranged as follows. In section 4.2, the issues of dynamic radio resource management are discussed. Also, the preliminary simulation results are presented. In section 4.3, the system model and the problem of dynamic cell configuration
are described. In section 4.4, the proposed dynamic cell configuration scheme is presented.
Simulation results are presented and discussed in section 4.5. Finally, section 4.6 concludes this chapter.
pilot traffic
(a). Fixed pilot scheme
CPICH DCH
pilot
traffic pilot
min max
traffic Downlink
Maximum Transmission Power
Free
(b). Dynamic pilot scheme
Free
Figure 4.1: Diagram of total power allocation of the base station in downlink CDMA systems with (a) fixed pilot and (b) dynamic pilot schemes.