slot slot slot … slot slot
Data subframe
slot slot slot … slot slotFrame n-1 Frame n Frame n+1
NENT NCFG NCFG … NCFG
Network Control subframe
CSCH CSCF CSCF … DSCH CSCH CSCF CSCF … DSCH
Schedule Control subframe
Frame n-1 Frame n Frame n+1
NENT NCFG NCFG … NCFG
Network Control subframe
CSCH CSCF CSCF … DSCH CSCH CSCF CSCF … DSCH
Schedule Control subframe
Data subframe
slot slot slot … slot slotData subframe
slot slot slot … slot slot Fig. 2. Frame structures of IEEE 802.16 mesh mode.
Requester Granter
Fig. 3. Three-way handshake.
3 However, since the flow that comes in and leaves a node shares the bandwidth. Equation (1) should be divided by two to represent the available bandwidth. Therefore, the end-to-end available bandwidth is:
2
By using (2), we define our SWEB metrics for all potential paths as: The path with the largest path metric will be chosen.V. ADMISSION CONTROL ALGORITHM:TAC
Our Token bucket-based Admission Control (TAC) has two essential parts. First, the bandwidth used by a connection must be estimated well. Second, the bandwidth estimation is used for implementing the admission control algorithm.
A. Bandwidth Estimation
If all the connections are under the control of token bucket mechanism, the bandwidth used with a time frame can be estimated as:
f
with a connection i, respectively. f is the frame length. However, (3) is over-estimated since the transmission burst does not happen in every time frame. To better estimate the bandwidth, consider the scenario in figure 4.Let the hop count and transmission deadline of the flow in Fig. 6 is 3 and 7f, respectively. Assume that the transmission burst occurs in time interval [t+5f, t+6f] and tokens stored in the
bucket are completely consumed. In order to satisfy the delay requirement, these bi bits of data must be sent in [t+9f, t+10f] at latest. Therefore, the frames from t+6f to t+10f can be used for sharing the bi bits, as in figure 5.
Generally speaking, in order to meet the delay requirement, di,
of real-time traffics, packets generated at time t
have to be sent after mi frames after t, whereTherefore, the maximum volume of data that can be sent in any given frame is:
i
B. Admission Control
We use the above-mentioned bandwidth estimation to implement the TAC algorithm. In TAC algorithm, the minimum usage of timeslots by each connection is defined.
They are: CBR_min, VBR_min and BE_min. When a station receives a MSH_DSCH:Request, it examines whether the current usage of each class exceeds their minimum usage or not.
If it is, the new-coming flow will be marked as downgradedٛ
flows. If a MSH-DSCH:Request comes in, the downgradedٛ
flows have bigger possibilities to be preempted. On the other
t+6f
Drop Precedence (2 bits)
Drop Precedence (2 bits) Reliability
(1 bit) Priority
(3 bits)
4 hand, if the current usage does not exceed its minimum usage,
the flow will not be downgrade and have bigger change to preempt other downgraded flows.
Since the service levels in IEEE 802.16 mesh mode are identified in the fields of CID (Connection Identifiers), we have the QoS mapping in Table I. With the mappings in table I, the down-graded flows can be marked. And by this information, we develop our TAC algorithm as follows:
1.) A new flow with its BW_req (Bandwidth request) in the unit of data timeslots. And set BW_avail as the total empty slot number. (BW_avail stands for available bandwidth) 2.) The station that handles the request checks if the
BW_req<BW_avail or not. If yes, go to step 3. Or else, go
to step 4.3.) The station determines to downgrade the flow or not, by comparing the current usage and the minimum usage of the traffic class.
4.) The station checks if the current usage exceeds the minimum usage of the traffic class. If yes, the flow shall be rejected. Or else, go to step 5.
5.) Check the timeslots used by downgraded flows in the order of BE_DG, VBR_DG, and CBR_DG. If there is no such timeslots, the request is rejected. Or else, set this timeslots empty, which means to preempt this timeslots. Updating the value of BW_avail. Go to step 2.
VI. SIMULATION RESULTS
The simulations are conducted in a 16-node topology, and the simulation area is a 4 km * 4 km square. The radio range is set as 1.5 km in radius. The frame length is chosen to be 8 ms.
In the simulations, QPSK is chosen to be the modulation method. The details of QPSK are given in table II.
The data rate of the CBR traffic is 64 kbps, with the 960-bit packet size in the packet interval of 15 ms. The VBR traffic is sending at the average speed of 400 kbps. The mean packet size is 16000 bits sending at the interval of 40 ms. The packet size of BE traffics is 8000 bits and is sent every frame (8 ms).
A. Routing
The proposed SWEB is compared with the ETX [7] and the shortest path. The performance and delay of VBR traffics are compared across all three different metrics. The performance is given in figure 6. And figure 7 shows the delay.
As shown in the figure 6 and 7, when number of flows is reaching 25, some VBR flows are preempted by CBR flows. By simulation results, we claim that SWEB is a compromise of
delay and throughput. But in figure 8, we can find that SWEB has best performance in jitter of real-time packets.
B. Admission Control
In TAC algorithm, the minimum usage of each traffic class must be set. In the simulations, the CBR_min, VBR_min and
BE_min are set as 10, 40 and 75 timeslots, respectively. Also,
the parameters of token bucket are shown in table III.TABLEII THE PARAMETERS OF QPSK
QPSK coding rate 3/4
OFDM symbols in a frame 676 OFDM symbols in a control subframe 16
OFDM symbols in a data subframe 660 OFDM symbols in a timeslot 4
Number of data timeslots 165 Capacity of a timeslot 144 bytes
Throughput
0 1000 2000 3000 4000 5000 6000 7000 8000
5 10 15 20 25
number of flows
bps
ETX Shortest SWEB
Fig. 6. Throughput of VBR flows.
Avg. Delay
0 10 20 30 40 50 60 70
5 10 15 20 25
number of flows
ms
ETX Shortest SWEB
Fig. 7. Delay of VBR flows.
TABLEIII
TOKEN BUCKET MECHANISM PARAMETERS
Token rate (bytes / frame)
Bucket size (bytes)
Delay requirements
CBR 120 8 40 ms
VBR 1500 500 80 ms
BE 7500 250 --
Jitter
0 5 10 15 20 25 30 35
5 10 15 20 25
Number of flows
time (ms) ETX
shortest SWEB
Fig. 8. Jitter of VBR flows.
5
We compare the throughput in figure 9 and figure 10. In figure 9, BE traffics suffers from preemption from higher priority traffic class, therefore, receiving low throughput when network is heavily-loaded. By applying the CAC algorithm in figure 10, the BE flows has the guaranteed throughput by setting the minimum usage. The preemption occurs only in down-graded flows.
In figure 11 and figure 12, the statistics is gathered to discuss the percentage of real-time packets exceeds the delay requirements. As in figure 11, around 12% of VBR-packets exceed the delay requirements when the number of flow is 25.
However, in figure 12 it is reduced to around 7% for only VBR-downgraded flows. It can be expected that for all VBR flows (VBR and VBR-downgraded), the ratio would be lower than 7%.
VII. CONCLUSIONS
In this paper, we proposed a new routing metric, SWEB, and an admission control algorithm, TAC for IEEE 802.16 mesh networks. SWEB is applied in static routing environment and yields the good throughput, delay and jitter performance. The TAC algorithm prevents the starvation of low-priority traffic flows and guarantees the delay requirements of the real-time flows. By SWEB and TAC, a QoS-enabled network environment can be realized with IEEE 802.16 mesh mode in the MAC layer. Thus, end-users will have better experience and convenience in utilizing the networks.
REFERENCES
[1] IEEE, “IEEE Standard for Local and metropolitan area networks Part 16:
Air Interface for Fixed Broadband Wireless Access Systems”, IEEE standard, October 2004.
[2] Harish Shetiya and Vinod Sharma, "Algorithms for routing and centralized scheduling to provide QoS in IEEE 802.16 mesh networks", Proceedings of the 1st ACM workshop on Wireless multimedia networking and performance modeling ,WMuNeP '05. Pages: 140-149.
[3] Tzu-Chieh Tsai, Chi-Hong Jiang, and Chuang-Yin Wang, "CAC and Packet Scheduling Using Token Bucket for IEEE 802.16 Networks", in Journal of Communications (JCM, ISSN 1796-2021), Volume : 1 Issue : 2, 2006. Page(s):30-37. Academy Publisher.
[4] Fuqiang LIU, Zhihui ZENG, Jian TAO, Qing LI, and Zhangxi LIN,
"Achieving QoS for IEEE 802.16 in Mesh Mode",8th International Conference on Computer Science and Informatics, Salt Lake City, USA [5] Hung-Yu Wei, Samrat Ganguly, Rauf Izmailov, and Zygmunt J. Haas,
"Interference-Aware IEEE 802.16 WiMax Mesh Networks", in Proceedings of 61st IEEE Vehicular Technology Conference (VTC 2005 Spring).
[6] Min Cao, Qian Zhang, Xiaodong Wang, and Wenwu Zhu, "Modelling and Performance Analysis of the Distributed Scheduler in IEEE 802.16 Mesh Mode", Proceedings of the 6th ACM international symposium on Mobile ad hoc networking and computing
[7] Douglas S. J. De Couto, Daniel Aguayo, John Bicket , and Robert Morris,
“A High-Throughput Path Metric for Multi-Hop Wireless Routing”, ACM MobiCom ’03.
exceeds delay requirement
0 2 4 6 8 10 12 14
5 10 15 20 25
number of flows
percentage (%)
CBR VBR
Fig. 11. The ratio of the realtime packets that exceeds the delay requirements for the original IEEE 802.16 mesh mode.
(TAC) throughput
0 1000 2000 3000 4000 5000 6000 7000 8000
5 10 15 20 25
number of flows
bps
CBR VBR BE total
Fig. 10. Throughput when TAC is applied.
throughput
0 1000 2000 3000 4000 5000 6000 7000 8000
5 10 15 20 25
number of flows
bps
CBR VBR BE total
Fig. 9. Throughput for the original IEEE 802.16 mesh mode.
exceeds delay requirement
0 1 2 3 4 5 6 7
5 10 15 20 25
number of flows
percentage (%)
CBR_DG VBR_DG
Fig. 12. The ratio of the realtime packets that exceeds the delay requirements when TAC is applied.