There are two approaches to distributed implementation of dynamic carrier allocation (DCA) strategies: the update approach and the search approach. We first investigate the fundamental differences between the two approaches: two simple and representative schemes, namely the basic update scheme and the basic search scheme, are presented and compared. We argue that the update approach is more appropriate than the search approach for mobile cellular networks. Then we propose an advanced update scheme that is more efficient than the basic update scheme in terms of message complexity and carrier acquisition delay. The advanced update scheme can support a group of DCA strategies which require resource planning.

Priority-based dynamic carrier allocation strategies can be classified into three categories: static-priority, dynamic-priority, and hybrid-priority strategies. Strategies based on static priorities do not consider the local carrier reuse conditions, while dynamic-priority strategies do take these conditions into consideration. Intuitively, one would expect dynamic-priority strategies to perform better than static- and hybrid-priority strategies, but in the literature it is the other way around-existing dynamic-priority strategies are out-performed by some static- and hybrid-priority strategies. We propose a dynamic-priority strategy which is a significant improvement over all existing strategies. Under various traffic conditions, our simulation results indicated that the proposed strategy could reduce the call blocking/failure rate by a margin ranging from 15% to 95%.

Recently, DQDB (IEEE 802.6) MAN has been proposed as a component of Personal Communication Networks, in which base stations of wireless infrastructures are connected by a number of DQDBs which in turn are connected via bridges. We propose a protocol for call setup and path migration in a cluster of DQDBs. The protocol uses a link-state-like routing method for path selection and a source-routing-based scheme for path establishment. In addition, we propose a labeling scheme that makes it possible to carry the path information needed by the source routing protocol in a single 53-octet DQDB slot. Without such a labeling scheme, source routing would be inefficient for our purpose.

The DQDB (IEEE 882.6) MAN has been proposed as a component of the personal communication networks, in which the base stations of the wireless infrastructures are connected by a number of DQDBs which in turn are connected by bridges. We propose a protocol for call setup and path migration in a cluster of DQDBs. To pursue simplicity, efficiency, and speed in path setup, we use a link-state-like routing method for path selection, and a source-routing-based protocol for path establishment. The feasibility of the source routing protocol in a DQDB cluster is guaranteed by our novel labeling scheme. Each label represents the routing direction from one DQDB to another such that a path can be represented by a sequence of labels. Because of the locality property of the labels, a few bits are enough to denote the labels unambiguously. Using multi-bit labels instead of multi-byte physical addresses, a path can be specified in a 53-octet DQDB slot.