Effects of Traffic Statistics of Existing Secondary Users’

Figure 6.4 shows how the existing secondary connections’ traffic statistics, including the average service time E[X_s] and the arrival rate λ_s, affect the cumulative handoff delay of the newly arriving secondary user’s connection.

10 12 14 16 18 20 2

3 4 5 6 7 8 9 10

Average service time of

the newly arriving secondary connection (E[χ

_s

])

Average cumulative handoff delay (E[D])

Random Strategy

Throughput−based Strategy Greedy Strategy

DP−based Optimal Solution ES−based Solution

(a) (E[Xp⁽¹⁾], E[Xp⁽²⁾], E[Xp⁽³⁾], E[Xp⁽⁴⁾]) = (14, 15, 15, 15), and (E[Xs⁽¹⁾], E[Xs⁽²⁾], E[Xs⁽³⁾], E[Xs⁽⁴⁾]) = (10, 12, 14, 16).

10 12 14 16 18 20 2

4 6 8 10 12 14 16

Average service time of

the newly arriving secondary connection (E[

χ_s

])

Average cumulative handoff delay (E[D])

Random Strategy

Throughput−based Strategy Greedy Strategy

DP−based Optimal Solution ES−based Solution

(b) (E[Xp⁽¹⁾], E[Xp⁽²⁾], E[Xp⁽³⁾], E[Xp⁽⁴⁾]) = (10, 15, 20, 25), and (E[Xs⁽¹⁾], E[Xs⁽²⁾], E[Xs⁽³⁾], E[Xs⁽⁴⁾]) = (10, 10, 10, 10).

Figure 6.3: Effects of the newly arriving secondary user’s average service time E[χ_s] on the cumulative handoff delay for λ^(k)p = 0.02 and λ^(k)s = 0.01 when 1 ≤ k ≤ 4.

Assume that the service time χ_s of the newly arriving secondary user’s con-nection is geometrically distributed with E[χs] = 10, and the primary users have the similar traffic parameters in different channels. From Fig. 6.4(a) where E[Xs^(k)] = E[Xs] for 1 ≤ k ≤ 4, we observe the following:

• In the range of small E[X_s] (e.g., E[X_s] < 15), the cumulative handoff delay increases as E[Xs] increases for the random selection strategy, the greedy strategy, and the optimal solution.

• In the range of large E[Xs] (e.g., E[Xs] ≥ 15), the secondary user will experience long waiting time when it changes its operating channel ac-cording to (6.6). Hence, the greedy strategy and the optimal solution prefer staying on the current operating channel when interruptions oc-cur to reduce handoff delay. In this case, their average handoff delay for each handoff is a constant E[Y_p]. Thus, the average cumulative handoff delay is also a constant. However, the random strategy still selects to change channel for spectrum handoff sometimes. Hence, the cumulative handoff delay of the random strategy still increases as E[Xs] increases.

• Because the throughput-orientated strategy always selects channel 1 for the target channel, the corresponding average handoff delay is a constant E[Yp]. Hence, the throughput-orientated strategy results in the same average cumulative handoff delay for various E[X_s].

Note that the similar observations can be also found in Fig. 6.4(b), where λ^(k)s = λ_s for 1 ≤ k ≤ 4. When λ_s ≥ 0.02, the interrupted secondary users will always stay on the current operating channel for the greedy strategy and the optimal solution.

5 10 15 20 2

3 4 5 6 7 8 9

Average service time of the secondary connections (E[X

s

])

Average cumulative handoff delay (E[D])

Random Strategy

Throughput−based Strategy Greedy Strategy

DP−based Optimal Solution ES−based Solution

(a) Effect of the average service time E[Xs] on the cumulative handoff delay E[D], where (λ⁽¹⁾s , λ⁽²⁾s , λ⁽³⁾s , λ⁽⁴⁾s ) = (0.01, 0.015, 0.02, 0.025).

0.012 0.015 0.02 0.025 0.03 3

4 5 6 7 8 9

Arrival rate of the secondary connections (λ

_s

)

Average cumulative handoff delay (E[D])

Random Strategy

Throughput−based Strategy Greedy Strategy

DP−based Optimal Solution ES−based Solution

(b) Effect of the arrival rate λs on the cumulative handoff delay E[D], where (E[Xs⁽¹⁾], E[Xs⁽²⁾], E[Xs⁽³⁾], E[Xs⁽⁴⁾]) = (10, 12, 14, 16).

Figure 6.4: Effect of the average service time E[X_s] and the arrival rate λ_s of the secondary users’ connections on the cumulative handoff delay of the newly arriving secondary user’s connection for (λ⁽¹⁾p , λ⁽²⁾p , λ⁽³⁾p , λ⁽⁴⁾p ) = (0.019, 0.02, 0.02, 0.02) and E[Xp^(k)] = 15 when 1 ≤ k ≤ 4.

6.5.3 Effects of Traffic Statistics of Existing Primary Users’ Connections

Figure 6.5 shows the effects of the average service time E[Xp] and the arrival rate λ_p of the primary users’ connections on the cumulative handoff delay of the newly arriving secondary user’s connection. We consider that λ^(k)s = λ_s and E[Xs^(k)] = E[X_s] for 1 ≤ k ≤ 4 as well as the service time χ_s is geometrically distributed and E[χs] = 10. In Fig. 6.5(a), we assume that E[Xp^(k)] = E[X_p] for 1 ≤ k ≤ 4. We can find the following:

• For all methods, the cumulative handoff delay increases as E[Xp] in-creases because a larger value of E[X_p] results in heavier traffic load.

• For the throughput-orientated strategy, the greedy strategy, and the optimal solution, their cumulative handoff delay at various E[X_p] will ultimately converge to the same value as shown in the region of E[Xp] ≥ 13 in Fig. 6.5(a). In the region, the handoff delay is only related to the busy period E[Yp] and uncorrelated to the value of E[Xs] because the interrupted secondary users always stay on the current operating channel when E[Xp] ≥ 13.

Note that we can have the similar conclusions in Fig. 6.5(b), where λ^(k)p = λ_p for 1 ≤ k ≤ 4.

5 10 15 20 2

4 6 8 10 12 14

Average service time of the primary connections (E[X

p

])

Average cumulative handoff delay (E[D])

Random Strategy

Throughput−based Strategy Greedy Strategy

DP−based Optimal Solution ES−based Solution

(a) Effect of the average service time E[Xp] on the cumulative handoff delay E[D], where (λ⁽¹⁾p , λ⁽²⁾p , λ⁽³⁾p , λ⁽⁴⁾p ) = (0.02, 0.025, 0.03, 0.035).

0.010 0.015 0.02 0.025 0.03 1

2 3 4 5 6 7 8

Arrival rate of the primary connections (λ

_p

)

Average cumulative handoff delay (E[D])

Random Strategy

Throughput−based Strategy Greedy Strategy

DP−based Optimal Solution ES−based Solution

(b) Effect of the arrival rate λp on the cumulative handoff delay E[D], where (E[Xp⁽¹⁾], E[Xp⁽²⁾], E[Xp⁽³⁾], E[Xp⁽⁴⁾]) = (10, 12, 14, 16).

Figure 6.5: Effect of the average service time E[X_p] and the arrival rate λ_p of the primary users’ connections on the cumulative handoff delay of the newly arriving secondary user’s connection for λ^(k)s = 0.01 and E[Xs^(k)] = 15 when 1 ≤ k ≤ 4.

Chapter 7

Reactive Spectrum Handoff

As discussed in Chapter 5, spectrum handoff mechanisms can be categorized as either the proactive spectrum handoff or the reactive spectrum handoff schemes. In this chapter, we focus on the modeling technique and perfor-mance analysis for the reactive spectrum handoff scheme. Compared to the proactive spectrum handoff scheme that the preselected target channel may no longer be available at the instant that spectrum handoff procedures are ini-tiated, the reactive spectrum handoff may have shorter handoff delay because it can reliably find an idle channel through spectrum sensing. Nevertheless, the reactive spectrum handoff scheme needs the sensing time to search the idle channels. In addition, it also needs the handshaking time to achieve a consensus on the target channel between the transmitter and receiver of a secondary connection. Hence, one important issue for the reactive spectrum handoff scheme is to characterize the effects of the sensing time and the hand-shaking time on the handoff delay. Obviously, when the sensing time and the handshaking time is too large, the reactive spectrum handoff is worse than the proactive spectrum handoff in terms of the extended data delivery time.

The goal of this chapter is to investigate the effects of spectrum handoffs

on the channel utilization and the extended data delivery time of the sec-ondary users’ connections with various traffic arrival rates and service time distributions. We consider the three key design features for spectrum han-odff, consisting of (1) heterogeneous arrival rates of the primary users at different channels, where various channels have different traffic arrival rates of the primary users because these channels may belong to different primary system operators; (2) various arrival rates of the secondary users at different channels, where the arrival rates can be determined by the initial operating channel selection mechanisms [79]; and (3) handoff processing time, result-ing from the sensresult-ing time, the handshakresult-ing time, and the channel switchresult-ing time. How to model the channel utilization at each channel and the extended data delivery time in the context of multiple handoffs is challenging since the operating channels for multiple handoffs are selected according to the chan-nel occupancy states at the moments of link transitions. To the best of our knowledge, an analytical model for characterizing all the three features for multiple handoffs has rarely been seen in the literature. The contributions of this chapter are summarized in the following:

• First, The preemptive resume priority (PRP) M/G/1 queueing network model is proposed to characterize the channel usage behaviors of CR networks. Based on this queueing model, we can evaluate the channel utilizations of different channels under various traffic arrival rates and service time distributions.

• Next, a state diagram is developed to characterize the effect of multiple handoff delay on the extended data delivery time of the secondary con-nections. Then, we can evaluate how long the extended data delivery time is prolonged due to multiple spectrum hanodffs.

7.1 System Model

7.1.1 Assumptions

In this chapter, we consider the spectrum handoff protocol presented in [95].

When the spectrum handoff procedures are initiated, the secondary users must spend τ slots on spectrum sensing to find the idle channels. Note that if more than one channel is assessed as idle, the interrupted secondary user will randomly select one idle channel from all idle channels to be its target channel for spectrum handoff. Here, we assume that this random selection follows the uniform distribution. Furthermore, the interrupted secondary user will stay on the current operating channel if all channels are busy. Next, the handshaking time of t_h slots is spent in order to achieve a consensus on the target channel between the transmitter and the receiver of a secondary connection. Hence, when a secondary user changes its operating channel to another channel, the total processing time for executing spectrum handoff procedures is δ_c , τ + t_h+ t_s where t_s (slots) is the channel switching time.

On the other hand, if the secondary user stays on the current operating channel, the total processing time is δ_s , τ + t_h.

7.1.2 Illustrative Example of Reactive Multiple

在文檔中 以排隊理論為基礎對感知無線網路頻譜管理技術之研究 (頁 149-160)