**2.2 System model**

34 Mobile Networks

whether there are packets destined to it. Message packets are checked to determine whether the MSS should be woken up or not. IEEE 802.16e defined three types of Power-Saving Classes (PSCs) for connections with different characteristics, and each PSC is defined for a set of connections with common properties. A PSC is composed of interleaved listening windows and sleep windows. In PSC Type I, the sleep window is exponentially increased from a minimum value to a maximum value. This is typically done when the MSS is doing best-effort and non-real-time traffic. PSC Type II has a fixed-length sleep window and is used for UGS service. PSC Type III allows for a one-time sleep window and is typically used for multicast

There are many previous researches that have devoted their efforts to adapting the sleeping duration of IEEE 802.11 and IEEE 802.15 (Liao & Wang, 2008; Liu & Liu, 2003; Tseng et al., 2002; Ye et al., 2004; Zheng et al., 2005). However, due to lack of QoS requirements, the results of those searches cannot be applied to IEEE 802.16e directly. Several studies have been proposed to analyze the IEEE 802.16e's power while an MSS operates in the power-saving mode (Han & Choi, 2006; Lei & Tsang, 2006; Seo et al., 2004). Several studies (Fang et al., 2006; Huang et al., 2007; Jang et al., 2006; Tsao & Chen, 2008) investigated the power consumption issues of IEEE 802.16e and suggested algorithms to determine the sleep interval in order to improve energy efficiency. In (Jang et al., 2006), the length of sleeping period is adapted according to the traffic type. This scenario is valid only under one MSS, and the QoS delay constraint is not considered. In (Tsao & Chen, 2008), although the QoS delay constraints are considered, the scenario cannot consider the energy costs of status transition. In (Fang et al., 2006), a scheduling algorithm for multiple MSSs with QoS delay constraints is proposed. To save power, the algorithm grants a primary MSS the right to use the bandwidth in burst mode. Secondary MSSs are only given the necessary bandwidth to meet the requirements of QoS delay constraints. However, its benefit only exhibits when the total traffic loading of all MSSs is low. In (Huang et al., 2007), although the constant bit rate traffic with QoS delay constraint is

In this chapter, we propose a QoS guaranteed energy-efficient scheduling for IEEE 802.16e. We consider that delay and jitter types of QoS should be scheduled at the same time and integrate sleep duration in one MSS. The packets would be scheduled successively to reduce the number of status transitions under QoS requirements for delay and jitter. The proposed approaches not only minimize the power consumption of the MSS but also guarantee both

In this section, we first describe the basic idea of our algorithm for QoS guaranteed energyefficient scheduling. Second, we define the notations of our system model. Finally, we schedule packets in an MSS with our QoS guaranteed energy-efficient scheduling. Additionally, we consider the QoS requirements of jitter constraint to schedule the packets

The idea behind our proposed algorithm, called successive scheduling scheme (SSS), is to schedule the packet transmission in successively fashion with the minimal interval of listen

**2. The QoS guaranteed energy-efficient scheduling for IEEE 802.16e** 

traffic or management traffic when the MSS knows when the next traffic is expected.

considered, the scenario cannot consider the jitter constraint.

delay and jitter QoS of real-time connections.

and achieve the guarantees of transmissions.

**2.1 Basic idea** 

In this chapter, the centrally controlled IEEE 802.16e wireless network with a central BS and an MSS with multiple real-time connections is considered. The uplink and downlink channel is divided into fixed-size frames, and the frames are subdivided into fixed-size time slots. Both the energy consumption and the bandwidth are calculated in time slots. Different QoS parameters have been defined for various type of services in IEEE 802.16e, and all of them can be mapped into the minimum data rate requirements of the MSSs (Andrews et al., 2005). Therefore, we only apply the minimum data rate as the bandwidth requirement of QoS for each type of connection. Additionally, other QoS requirements such as the maximum latency and tolerated jitter would be considered in this chapter. The notations in this chapter are as follows: *Taw* is the total number of time slots in which an MSS stays in the awake state; *Tst* denotes the total number of status transitions of an MSS from the sleep state to the awake state; *Paw* stands for the average energy consumption of each time slot by an MSS in the awake state; *Pt* represents the average energy consumption of each status transition from the sleep state to the awake state in an MSS; *n* denotes the index of time slot in an MSS; *rn* stands for the data rate in which an MSS has been allocated by time slot *n*; min *Rn* stands for the minimum data rate that an MSS should receive in order to guarantee its service quality in time slot *n*. We assume that there is no energy consumed during the sleep period of an MSS. Thus, the energy consumed of an MSS is determined by the number of the time slots it stays in the awake state and the number of status transitions it has from the sleep state to the awake state. The overall energy consumed by an MSS during period *T*, denoted as *P*, can be represented as follows:

$$P = T\_{\text{aww}} \times P\_{\text{aw}} + T\_{\text{st}} \times P\_{\text{ft}} \tag{1}$$

The goal of the scheduling algorithm is to minimize the average energy consumed by an MSS during period *T*, while the QoS requirements such as minimum data rate, maximum delay constraint and tolerated jitter of an MSS must be guaranteed. Thus, we can minimize *P* by allocating the minimum time slots (*Taw*) to satisfy the minimum data rates ( min *Rn* ) and successively schedule the packets to reduce the status transitions (*Tst*). In order to acquire the optimal result, the power-saving scheduling algorithm should consider the properties of the QoS requirements. We discuss the solutions of previous studies and present our QoS guaranteed energy-efficient scheduling to acquire the optimal result in the next section.
