مدل سازی و تجزیه و تحلیل از دولت مدیریت تحرک خدمات تغییر بسته اطلاعاتی در خدمات عمومی بسته رادیویی

There has been an increasing demand for wireless data services due to the popularity of Internet services and circuit-switched (CS) systems are not appropriate for accommodating bursty data traffic. The wireless data services can be efficiently supported in the packet-switched (PS) system and General Packet Radio Service (GPRS) is a representative PS system which is being serviced widely. In GPRS, three mobile station (MS) mobility management states, i.e., idle, ready, and standby are defined in order to accommodate bursty traffic characteristics of data services, and thus, GPRS results in efficient management of radio resources and signaling networks. In order to analyze the performance of GPRS mobility management, we develop an analytical model to derive the steady-state probability of the MS states, which is essential in the performance analysis. The analytical model is validated by using a simulation model. The effect of various input parameters on the steady-state probability and the effect of variances of cell residence time, RA residence time, and packet transmission time are analyzed. Then, location update signaling and paging signaling loads are investigated based on the steady-state probability. Our study provides guideline for proper selection of PS system parameters and can be used to analyze the performance of mobility management schemes for PS systems.

Mobility management, which consists of location registration and call delivery, has been considered as one of the most important issues in mobile communication networks and there have been numerous studies in the literature [1], [2], [3], [4], [5], [6], [7] and [8]. The location registration notifies the network of the location of mobile stations (MSs), allowing the network to maintain the latest location information of the MSs. In call delivery, this location information is interrogated and the call is delivered to the called MS after paging. For both location registration and call delivery there is a tradeoff between the use of radio resources and signaling network resources. Most previous studies have focused on analysis of location registration signaling and paging signaling loads. Recently, there has been an increasing demand for wireless data services due to the popularity of Internet services [9]. Second generation (2G) mobile communication systems such as global system for mobile communications (GSM) do not meet this demand because data services are provided based on circuit-switched (CS) radio transmission, where a channel is exclusively allocated for a single user for the entire call duration, even if the user does not transfer data packets at the moment [9]. This results in highly inefficient use of resources for data services with bursty traffic characteristics. In order to overcome this problem, packet-switched (PS) systems have been proposed. In cellular systems, General Packet Radio Service (GPRS) [9], [10], [11], [12], [13] and [14] is a representative PS system which is being serviced widely. In GPRS, a channel is allocated when needed and is released immediately after the transmission of packets is completed and thus multiple users can share the same channel. In GPRS, MSs are defined to have three states: idle, ready, and standby. In the idle state, the MS is not reachable and the GPRS network does not have any location information. In the ready and the standby states, cell and routing area (RA) (i.e., a group of cells) information of the MS is managed, respectively. If the packet transmission rate is high and the MS is expected to receive packets in short intervals, the MS stays in the ready state. In the ready state, network can deliver packets to the MS without paging all cells in an RA. On the other hand, if the packet transmission rate is low, the MS stays in the standby state and does not update its location frequently. In the standby state, paging all cells in the RA is required to deliver packets to the MS. The GPRS state model accommodates the properties of PS data services, while GSM state model is appropriate only for CS voice services. In GSM, location update occurs as MS moves across location areas (LAs) and paging all cells in an LA is performed if any incoming call arrives. The GSM MS state, whether it is idle or busy, does not affect to obtain location update signaling and paging signaling loads. In GPRS, however, cell updates occur in the ready state and RA updates occur in the standby state, and paging is performed only in the standby state. Thus, signaling load by location update and paging depends on the probability of the ready and the standby states, and derivation of the steady-state probability of MS states is essential to analyze the performance of GPRS mobility management. The study of steady-state probability of MS states is also essential in the performance analysis of PS micro-mobility protocols which are considered as one of the promising candidate protocols for fourth generation (4G) IP-based mobile networks. In these protocols, two MS states for PS services are defined to accommodate bursty traffic characteristics of data services and the concept of paging area (PA) is proposed. For example, active and idle states are defined in P-MIP [15]. In the active state, registration occurs whenever the MS changes its cell. On the contrary, registration occurs only if the MS changes its PA in the idle state. If there is any incoming data for the idle MS, paging is performed to find the exact location of the called MS. In [16], the performance of the P-MIP was analyzed based on the steady-state analysis. In the IETF seamoby working group [17], [18], the concept of active and dormant modes have been proposed. Similar to the P-MIP protocol, MS performs cell-based location update in the active mode and PA-based location update in the dormant mode. Also, in HAWAII [19], active and standby states are defined and paging is needed in the standby state, where PA-based location update occurs. As described above, most PS systems have two distinct MS mobility management states for the registered MS (i.e., ready and standby states for GPRS). In one state (i.e., ready), MS updates its location upon every cell change. In the other state (i.e., standby), MS updates its location when the MS changes its RA or PA and paging is needed in this state in order to deliver incoming packets. The state transition from active to idle states is controlled by a timer (i.e., active timer in GPRS). In order to analyze the performance of these PS systems, the derivation of the probability of the MS states is essential. Since the analysis of the state management of CS voice services [20], [21], [22] and [23] is not valid for PS services, a study on the state management for PS data services is required. In our previous study [24], we derived the steady-state probability of GPRS MS states based on simple exponential distribution assumptions on cell residence time, RA residence time, and packet transmission time. In this paper, as an extension of our previous study, we develop an analytical model to derive the steady-state probability of the GPRS MS states based on more practical distributions, i.e., Erlang distribution on cell and RA residence times and Gamma distribution on packet transmission time, and the derived analytical results are validated by using the simulation results. Then, the effect of various input parameters on the steady state probability and the effect of variances of cell residence time, RA residence time, and packet transmission time are analyzed. Finally, location update signaling and paging signaling loads are investigated based on the steady-state probability. These analysis results provide guideline for proper selection of PS system parameters and can be used to analyze the performance of any mobility management schemes for PS systems. Although we are only concerned with GPRS system in this paper, the analytical model itself can be applied to other PS systems easily, which has similar MS mobility management states. This paper is organized as follows: Section 2 presents a GPRS network architecture and MS state management. The steady-state probability of the MS states is then derived in Section 3. In Section 4 numerical examples are provided in order to show the effects of various input parameters on the steady-state probability, and the signaling load of location update and paging is also analyzed. Finally, conclusions are presented in Section 5.

In this paper, we proposed an analytical model to derive the steady-state probability of GPRS MS states. The MS state transition behavior is modeled based on a modified GPRS state transition diagram and the steady state probability of MS states is derived using a semi-Markov process approach. In numerical examples, the analytical model was validated by using a discrete-event simulation model and the effect of various parameters on the steady-state probability is analyzed. It was demonstrated that the steady-state probability of ready state increases as ready timer, mobility of MS, and packet arrival rate increase. We also investigated the effect of variances of cell residence time, RA residence time, and packet transmission time. The results indicate that the steady-state probability of ready state decreases as the variances of cell and RA residence time decreases. On the other hand, the steady-state probability is insensitive to the variance of the packet transmission time is negligible. Finally, we evaluated the tradeoff between location update signaling and paging signaling loads. These results demonstrate that careful selection of system parameters is required for efficiently accommodating PS services and provide guideline for proper selection of PS system parameters. The results also can be used to analyze the performance of mobility management schemes for PS systems. Although we were only concerned with GPRS system in this paper, the analytical model itself can be applied to other PS systems easily, which has similar MS mobility management states.