Presented by:Ayush Gupta
Mobility Modeling During Rush Hour Traffic
User mobility is one of the most important factors affecting the network performance in a cellular network. Increased mobility results in more location updates, handovers and hence an increase hi the number of messages exchanged between various entities in the system. This signaling not only puts extra load on the radio interface but also on the infrastructure equipment. Thus, it is required to minimize the rate of occurrence of these mobility related events by optimally dividing the system into areas that have a minimum exchange of traffic. This report presents a mobility model that can be used to predict the transient behavior of the traffic on the routes connecting these areas in a given network during rush hours when the capability of the network to handle mobility related signaling traffic is put to a severe test. Finally we will study the practical implication of model formed in GPRS technology.
There are two types of mobility supported in Personal Communication Services-Terminal mo¬bility Personal mobility. Personal mobility refers to the network's ability to recognize a user irrespective of his position as well as the terminal. In terminal mobility, however, the terminal follows some procedure to update the network of its location even if it is not currently in use. Here we deal only with terminal mobility problem in cellular networks. Mobility of subscribers presents several technical problems Cellular networks use a combination of location updates and paging to provide roaming for its users. When a mobile enters a new location area, it sends a location update message to register itself with that area. When a call arrives for the mobile, all the cells in the location area are paged. Both location updates and paging contribute to the signaling traffic in the network. In GSM systems. SDCCH (Stand-alone Dedicated Con¬trol Channel) are used for location updates. If location areas are not chosen optimally, then the number of SDCCH channels to be dimensioned to support communication establishment attempts would bo increased, In addition to the radio resources, location updates also trigger an exchange of messages in the network infrastructure.
In this report I present a mobility model for modeling rush hour traffic in a city area. This model can be used to obtain mobility measures for routes with high traffic density during rush hour. These mobility measures along with paging measures can be used to solve a graph-partitioning problem to obtain optimal/near optimal LA design.
Further we extend our report to a practical level of design implementation MLST formation to enable the equivalent accounting of potential handoffs at a core link, and multiple dynamic guard bandwidth scheme with quantized guard bandwidths reserved on individual links over the GPRS core network.
Figure shows a typical city area in which several area zones are connected by high capacity routes. Area zones could be of any of the four types - city center, urban, suburban, and rural. High capacity routes represent the most frequently selected streets for support of movement between different area zones. Optimal partitioning of a wireless network into contiguous areas viz. location/registration areas requires the knowledge of traffic distribution at various routes as a function of time. Also, it is expected that the cost of a partition would be maximum during rush hour periods, when exchange of traffic between various area zones is maximum. This can be seen as a graph-partitioning problem with each area zone representing a node and the high capacity routes representing the edges connecting these nodes. Each node and edge is assigned a weight. In ease of edges, this weight represents the cost of that edge being intersected by a partition (mobility measure), while in case of nodes it signifies the terminating BHC'A at the cell or area-zone represented by the node (paging measure}. Given the set of nodes (vertices), V and the set of edges E. a forma! definition of the problem is as follows: Let G = (V, E) be a graph representing nodes from V. connected by edges in E. The graph is to be partitioned into disjoint subsets of nodes so that the sum of the weights on inter subset edges is minimized, while ensuring that the sum of node weights in a subset is at most M.