Scheduling and Call Admission Control (CAC) in IEEE Mesh Networks

Scheduling and Call Admission Control (CAC) in IEEE 802.16 Mesh Networks M.Tech. Project Second Stage Report Submitted in partial fulfillment of the r...
Author: Alfred Riley
7 downloads 2 Views 852KB Size
Scheduling and Call Admission Control (CAC) in IEEE 802.16 Mesh Networks M.Tech. Project Second Stage Report Submitted in partial fulfillment of the requirements for the degree of Master of Technology by Jeevan B. Chalke Roll No: 06329011 under the guidance of

a

Prof. Anirudha Sahoo

Department of Computer Science and Engineering Indian Institute of Technology, Bombay Mumbai January 2008

Acknowledgements I would like to thank my guide Prof. Anirudha Sahoo, for his invaluable help and guidance during the course of my MTP II stage. I also thank my friends Janak and Zahir for their support. I am highly indebted to all for constantly encouraging me by giving their criticisms on my work.

Declaration This report is based on the input from sources that I have found on the Internet. I acknowledge the use of these sources carefully and diligently. Some of the figures used in the report have been taken from these sources. The source for each figure has been mentioned along with the figure. A list of the sources that were referred for the creation of the report appears in the bibliography.

Jeevan B. Chalke CSE, IIT Bombay January 2008

Abstract The IEEE 802.16 - WiMax provides a mechanism for deploying broadband wireless networks in geographically large areas. It supports three modes: Point to Point (P2P), Point to Multi-Point (PMP) and Mesh. The MAC and physical layer of WiMax is carefully designed to support several real time applications with quality of service (QoS) guarantees. In the absence of proper scheduling and admission control, the system cannot provide promised QoS guarantees. Unfortunately the IEEE 802.16 standard does not specify any scheduling or admission control mechanism in PMP and Mesh mode. In mesh mode, the network topology is a tree rooted at base station and the problem is to determine the routing, link scheduling and call admission control for the tree. In this report we present an efficient centralized scheduling scheme for WiMax mesh networks on top of the routing algorithm designed. An efficient centralized algorithm that we have proposed not only increases the overall throughput of the system but also admits more number of connections.

Keywords:

IEEE 802.16 (WiMax) Networks, Mesh Networks, Quality of Service (QoS),

Scheduling, Routing, Call Admission Control (CAC)

Contents 1

Introduction

3

2

Overview of 802.16 mesh mode

5

2.1

Mesh mode frame structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5

2.2

Centralized Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6

3

4

5

Problem Definition

9

3.1

Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9

3.2

Problem Statement and Description . . . . . . . . . . . . . . . . . . . . . . . .

9

Routing

11

4.1

Building Routing Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4.2

Parent Selection Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

4.3

Parent Selection Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

An Efficient Centralized Scheduling

15

5.1

Parallel Transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

5.2

Admitting More Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

6

Simulation Results

22

7

Proposed Plan

25

8

Conclusion and Future Work

26

Bibliography

27

1

List of Figures 2.1

A typical mesh network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6

2.2

An example of an IEEE 802.16 based mesh network (source: [2]) . . . . . . . .

6

2.3

IEEE 802.16 mesh frame structure (source: [1]) . . . . . . . . . . . . . . . . . .

7

2.4

Flow usage example (source: [1]) . . . . . . . . . . . . . . . . . . . . . . . . .

8

4.1

New node entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

4.2

Parent selection algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

5.1

Parallel transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

5.2

Number of frames required to transmit packet at last hop . . . . . . . . . . . . . 16

5.3

Sequence of requests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5.4

A scheduling tree example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5.5

Slot allocation for tree shown in Figure 5.4 . . . . . . . . . . . . . . . . . . . . . 18

5.6

A scheduling tree with more number of connections, slots = 20 . . . . . . . . . . 19

5.7

Admitted requests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

5.8

Slot allocation for tree shown in Figure 5.6, slots = 20 . . . . . . . . . . . . . . . 20

5.9

Call admission control algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . 21

6.1

Adding new node in a routing tree . . . . . . . . . . . . . . . . . . . . . . . . . 22

6.2

Routing table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

6.3

Routing tree generated and used in scheduling . . . . . . . . . . . . . . . . . . . 23

6.4

Admitted connections with slots/request = 5 . . . . . . . . . . . . . . . . . . . . 24

6.5

Admitted connections with slots/request = 10 . . . . . . . . . . . . . . . . . . . 24

6.6

Admitted connections with slots/request = 25 . . . . . . . . . . . . . . . . . . . 24

2

Chapter 1 Introduction The IEEE 802.16 protocol for wireless metropolitan area networks (WMAN) has been recently standardized to meet the needs of wireless broadband access. The IEEE 802.16d [1] supports point-to-point, point-to-multipoint and mesh topology and provides a scalable solution for extension of a fiber optic backbone. A WiMax base station can offer greater coverage area around five miles with line-of-sight (LOS) transmission within bandwidth of up to 70 Mbps. WiMax is suitable for many neighborhoods that are too remote to receive Internet access via cable or DSL, and for any place where the cost of laying or upgrading landline to broadband access is expensive. When backhaul based WiMax is deployed in mesh mode, it not only increases the wireless coverage but also provides features like lower backhaul deployment cost, rapid building, easy deployment, robustness and re-configurability. The 802.16 mesh networks are mainly used to provide cost effective Internet access to sparsely populated areas. Thus the network topology is a tree routed at the base station and the problem is to determine the routing, link scheduling and call admission control for the tree, either jointly or separately. In community wireless networks, most of the nodes are either stationary or minimally mobile. Hence the focus of routing algorithm is on improving the network capacity of the performance of the individual transfers, instead of coping with mobility or minimizing power usage. Scheduling and call admission control plays an important role in increasing the spatial reuse, achieving high throughput and providing fair access to the subscriber stations. The call admission control analyzes the request, and determines the acceptance or rejection of request, and if accepted, its QoS attributes are registered in its service flow database. Scheduling then uses this service flow database and routing algorithm to schedule the connection in a frame. 3

In this report we present the routing algorithm and an efficient centralized scheduling scheme which makes use of parallel transmission to increase the spatial reuse and hence increases the overall system throughput. The scheduling presented admits more number of connections by reserving the bandwidth or slots only at the required hop. Our simulation results show that there is significant increase in number of connections admitted. The report is organized as follows; in Chapter 2, we discuss the overview of IEEE 802.16 mesh mode. Chapter 3 will then present the problem definition. How to build a routing tree, what are various parent selection criteria are presented in Chapter 4 called routing. In Chapter 5 we present our an efficient centralized scheduling scheme. We present our simulation result in Chapter 6 followed by proposed plan for MTP stage III in chapter 7. Finally we present our conclusion and future work.

4

Chapter 2 Overview of 802.16 mesh mode The IEEE 802.16 provides both PMP and mesh topologies. The main difference between PMP and mesh mode is that in PMP mode, traffic only occurs between base station (BS) and subscriber stations (SSs) where as in mesh mode; traffic can be routed through other SSs as well. As shown in Figure 2.1, the overall area is divided into meshes and managed by a single node called as mesh base station (MBS). It serves as an interface for WiMax base mesh to the external network. A transmission can occur between two SSs within a mesh or within two meshes. Transmission within a single mesh may or may not involve the MBS where as transmission between SSs in two different meshes must route through the MBS. In the later case, SS routes its packets to its MBS via several SSs in between, MBS then routes it to BS (through backhaul link) which forwards the packet to the destination SS’s MBS. Finally the MBS of the destined SS routes the packet to SS via several SSs in the route. For example, in Figure 2.1 packet from SS3 to SS4 will take path as: SS3 - SS2 - MBS (Mesh 1) - BS - MBS (Mesh 2) - SS4.

2.1

Mesh mode frame structure

The IEEE 802.16 mesh mode MAC supports both centralized and distributed scheduling. Centralized scheme is widely used to establish the high speed broadband mesh connection, in which MBS coordinates the radio resource allocation within a mesh network. Contrary to PMP mode, there are no separate uplink and downlink subframes in mesh mode. The mesh mode only supports TDMA - Time Division Multiple Access for channel access among MBS and SSs. A mesh frame consists of control and data subframe. The control subframe serves two functions: network control and schedule control. The data subframe is shared between centralized and 5

Figure 2.1: A typical mesh network

Figure 2.2: An example of an IEEE 802.16 based mesh network (source: [2])

distribute scheduling. Figure 2.3 shows the IEEE 802.16 mesh frame structure. In a network control subframe, mesh network configuration (MSH-NCFG) and mesh network entry (MSHNENT) packets provide basic level of communication for nodes to exchange network configuration information. In a schedule control subframe, mesh centralized scheduling (MSHCSCH) and mesh centralized scheduling configuration (MSHCSCF) packets are used for transmission bursts corresponding to centralized messages, and the rest are allocated to transmission bursts containing mesh distributed scheduling (MSH-DSCH) packets for distributed scheduling. The data subframe consists of minislots. Scheduled allocation may consist of one or more minislots.

2.2

Centralized Scheduling

In mesh topology, unlike PMP mode there are no clearly separated downlink and uplink subframes. Each station can create a direct communications link to a number of other stations in the network instead of communicating with only the MBS. In mesh centralized scheduling, the MBS perform most basic functions as BS does in PMP. The key difference is that in mesh mode all the SSs may have direct links with other SSs. Communication in all these links should be controlled by centralized algorithm. In this section we see how centralized scheduling is proposed in IEEE 802.16d standard [1]. The schedule using centralized scheduling is determined in more of a centralized manner by MBS. The scheduled transmissions for SSs should be defined by the MBS. MBS determines the

6

Figure 2.3: IEEE 802.16 mesh frame structure (source: [1])

flow assignments from the requests (DSA or bandwidth) from SSs. Thus MBS acts like a BS in PMP mode except that not all the SSs have to be directly connected to the MBS, and assignments determined by the BS extends to all SSs which are not directly connected to the MBS. The MBS should gather resource requests from all the SSs. It should then determine the amount of granted resources for each link in the network both in downlink and uplink, and communicate these grants to all the SSs through scheduling. Centralized scheduling ensures that transmissions are coordinated and are collision free over the links in the routing tree to and from the MBS. A simple example of the use of the centralized scheduling flow mechanism is shown in Figure 2.4. In Figure 2.4 note that the intermediate parent sends aggregate flow request (request from SS3 + SS4 + its own) to the MBS irrespective of the level (or hop) at which they are being utilized. A more concrete example depicting the same with absolute slot requests is provided in Chapter 5. MBS periodically transmits the new schedule after some number of frames to all the nodes. We will refer to this period as the scheduling period. For example, after every 10 frames, the schedule is transmitted. Therefore we can say scheduling period is of 10 frames. All the nodes

7

Figure 2.4: Flow usage example (source: [1])

use this newly arrived schedule to transmit in their respective frame. MBS uses routing tree to create the schedule and follows the following steps for transmission. 1. The MBS transmits first in a new frame. 2. Then in next frame, all eligible children of the MBS (i.e., nodes with hop count equals 1), transmit. 3. Then, the eligible children of the nodes from step 2 (i.e., nodes with hop count equals 2), transmit. 4. The process continues until all eligible nodes in the routing tree have transmitted. In this way if there are maximum n hops in the network, each node gets chance to transmit after every n frames.

8

Chapter 3 Problem Definition 3.1

Motivation

WiMax is a newer technology standard which offer greater coverage and bandwidth upto 70 Mbps. Mesh topology is similar to a multi-hop ad-hoc network. The communication can happen between two SSs without involving MBS as explained in Chapter 2. Implementing mesh topology is kept optional in the IEEE 802.16 standard and hence there is very less work done in the area of WiMax mesh. Scheduling and call admission control are important modules in providing QoS guarantees. Also, IEEE 802.16d standard does not specify any algorithm for scheduling and call admission control (CAC). Due to its multi-hop nature, various algorithms designed for scheduling and CAC in PMP mode cannot be directly applied on such mesh network. Thus we need an efficient scheduling and CAC mechanism for such a multi-hop mesh network.

3.2

Problem Statement and Description

The goal of the project is to design an efficient scheduling and CAC mechanism for IEEE 802.16 mesh networks. Since WiMax mesh is a multi-hop network, we have to decide upon the path from each SSs to the MBS along which the transmission will take place and hence we need to design the routing algorithm as well. Scheduling and CAC will then use this routing tree for designing the schedule and admitting flows/connections respectively. 9

By efficient scheduling we mean the scheduling algorithm and CAC should 1. Admit more number of connections 2. Increase the overall system throughput 3. Provide QoS guarantees for various service flows like UGS, rtps, and nrtps Chapter 5 will explain in more detail the scheduling mechanism proposed for admitting more number of connections, how it utilizes spatial reuse and parallel transmission for increasing the overall system throughput. Providing delay guarantees is kept as future work and will be completed in third stage of the project.

10

Chapter 4 Routing We know routing is closely coupled with scheduling. Scheduling uses a tree rooted at the MBS defined by the routing algorithm for generating a schedule map. Routing ensures that there is a unique path from every SS to the MBS which might go through several SSs. Routing algorithm should design such a tree and assign the parent to the newly arrived node. Here in this chapter we will present how the routing tree is built and some parent selection criteria to choose a better node as its parent if there are multiple potential parents.

4.1

Building Routing Tree

To build a routing tree, initially we start with a tree having a single node in the network i.e. MBS, root of a tree. Then we will add nodes to the tree in one by one fashion. These nodes are assumed to be new node arrivals in a network and hence follow following steps to determine the parent. 1. New node sends (broadcast) NEW-ENTRY packets after listening the beacon. The message contains information such as MAC address. 2. All neighbors hear this packet and reply with following information: hop count, average packets, number of children and RSSI. 3. This new node waits for 3 consecutive frames to gather such information from all its neighbors. Since a new node may arrive at any level (hop), when it advertises its existence, nodes at same level (hop), nodes at one level above and one level below will also listen to its advertisement and hence send their response in 3 consecutive frames in the next scheduling interval. 11

4. The new node now has multiple responses from various neighbors. After that it will run parent selection algorithm to decide upon its parent. 5. It now sends an acknowledgement to the parent selected by the algorithm. This is a unicast message since the new node knows its parent now. 6. The parent then forwards this information to the MBS to update its routing tree. 7. MBS will now update its routing tree and also know the path from this newly arrived node to all other nodes in a network. All these steps and message flows for establishing a link between newly arrived node and its parent are presented graphically in Figure 4.1 shown below.

Figure 4.1: New node entry

4.2

Parent Selection Criteria

Nearby nodes which hear the NEW-ENTRY packet send the response to the newly arrived node. Hence there might be several multiple responses at the new node. There are various different metrics or criteria on which parent selection algorithm decides the parent from these multiple responses. Here in this section we present some possible criteria for parent selection.

12

Hop Count Hop count criterion provides minimum hop count routing. The link quality for this criterion is a binary concept; either the link exists or it doesn’t. The primary advantage of this criterion is its simplicity. It is easy to compute the minimum hop from the responses and always ensures the shortest path from MBS. Moreover, computing the hop count requires no additional measurements like signal strength or average packets at the node. Also another advantage of minimum hop count is that it ensures chain topology. Nodes at hop 1 and hop 4 are far away from each other such that both are in non-interfering range and can start communication at the same time.

RSSI Received Signal Strength Indication (RSSI) is a measurement of the power present in the received radio signal. More the RSSI value, better is the link between two nodes. In our simulation we assume a Cisco Systems card which returns a RSSI in between 0 and 100 [15].

Average Packets at Node Average packets at each node indicates the overall traffic routing through the node. More the traffic, more processing time spent at the node. A heavily loaded node may become a bottleneck for a network.

Number of Children Number of children also affects the total load on a node. More number of children increases the processing time at a node. A single link failure between a node and its parent will automatically disconnect all its children from the network. Hence assigning a parent with more number of children is not a good idea.

4.3

Parent Selection Algorithm

Figure 4.2 presents the proposed parent selection algorithm. As explained in previous section minimum hop is a first level criterion for selecting the parent whereas we used RSSI as our second

13

1: Find minimum hop from all the responses, let it be minHop 2: for Each response do 3:

if hop equals minHop then Add in a list of minHopList

4: 5:

else Reject the response

6: 7:

end if

8: end for 9: if There are only one such node then 10:

Select it as parent

11: else 12:

Select parent with maximum RSSI from minHopList

13: end if

Figure 4.2: Parent selection algorithm level selection criterion. Further study on other two criteria is in progress and these three together with some weight will be the second level criteria.

14

Chapter 5 An Efficient Centralized Scheduling Centralized scheduling scheme mentioned in the standard does not utilize spatial reuse. It also reserves the required number of slots at all the hops in the path irrespective of the destination. This leads to poor throughput, poor slot utilization and admitting less number of connections in the system. In this chapter we present a parallel transmission mechanism to increase the overall system throughput and a mechanism for admitting more number of connections with suitable examples.

5.1

Parallel Transmission

Minimum hop as a first level criterion for selecting the parent in building a routing tree encourages to use of the parallel transmission. Due to this minimum hop criterion nodes at hops h, h + 3, h + 6, ... can transmit or receive at the same time since they are not in the interference range of each other. This leads to a 3-hop parallelism. As shown in Figure 5.1, while communication on hop 1 is going on, simultaneous communication at hop 4 is possible. For example, when node 0 transmits1 , node 1 will receive data from it. We assume that node cannot transmit and receive at the same time and thus node 1 cannot able to transmit when it receives data from node 0. Also node 2 cannot initiate its transmission since it affects the reception at node 1. At the same time, node 3 can initiate its transmission at hop 4 leading to 3-hop parallelism. Similarly in next frame both node 1 and node 4 can initiate their communication i.e. on hop 2 and hop 5 simultaneously. 1

Since there are no separate downlink and uplink subframes, a node can transmit or receive in any slot. Hence

by transmit actually we want to say transmit or receive and by receive we mean receive or transmit

15

Figure 5.1: Parallel transmission

Due to 3-hop parallelism every node gets an opportunity to transmit their data after every 3 frames rather than after n frames where n is the number of hops in the network. Thus if there is a packet at the last hop, number of frames required (F ) for that packet to reach MBS is given by F = 2(HC − 1) + T XOP

(5.1)

Where, HC is a hop count and T XOP is the transmission opportunity. Transmission opportunity here means when the node at hop h gets a frame to transmit. Since we are using 3-hop parallelism it will be either 1, 2 or 3 and given by T XOP = (HC − 1)%3 + 1

(5.2)

Figure 5.2: Number of frames required to transmit packet at last hop

As shown in Figure 5.2 we compared our proposed 3-hop parallelism with the scheme mentioned in the IEEE 802.16d standard. We observe that in the scheme explained in the standard, 16

as number of hops increase in the network the number of frames required to transmit the packet at last hop increases exponentially, whereas our 3-hop mechanism increases linearly.

5.2

Admitting More Connections

As explained in Chapter 2, each node sends an aggregate slot request to the MBS. Figures 5.3, 5.4, and 5.5 show an example of such an aggregate slot request.

Figure 5.3: Sequence of requests

Figure 5.4: A scheduling tree example

17

Figure 5.3 represents the sequence of request made by each SS. Figure 5.4 shows corresponding scheduling tree with downlink/uplink request on each node, whereas Figure 5.5 shows the frames at each hop with slot allocation ordered by arrival of request. Here we assume sufficient slots (i.e. 32) and hence all requests are admitted.

Figure 5.5: Slot allocation for tree shown in Figure 5.4

Now let us consider that each frame has 20 slots. With this constraint it is clear that sufficient slots are not available to admit all requests. If we admit the request one at a time only the first 8 requests will get admitted and the remaining will be rejected by the system. But if we observe carefully, in request 6, SS 8 wants to communicate with SS 7 and thus there will not be any slot utilization on hop 1 hence wasting 6 slots on hop 1 which would otherwise be used by request 10. With such allocation of slots, as the number of hops increase the slot utilization decreases, see Figure 5.5. To reduce such under utilization of slots on higher level and to minimize the wastage of slots, we enforced the following constraints in our scheduling 1. Each node should mention the destination node to which it wants to communicate in a DSA request. Destination equal to 0 indicates data transfer outside the mesh. 2. Reserve slot only at required hop. Here are the steps followed in our proposed scheduling algorithm 1. A node wishing to transmit or receive will send a DSA/DSC/DSD request stating clearly the required bandwidth/slots and other QoS details. It will also mention the destination node in a request. 18

2. All intermediate parents, if any, will simply forward this request to the MBS. 3. When MBS receives such a request it is routed through a CAC module. 4. CAC looks for priority if there are multiple requests in a queue and processes the requests as per their priority. CAC module admits the flow if and only if there are sufficient slots on all the required hops. 5. Scheduler will then generate a map stating the SSID, starting slot number and the number of slots assigned for all admitted requests. 6. After receiving the schedule, all the 3-hop nodes will start their transmission at the same time from next frame. Figure 5.9 presents the call admission control algorithm. First, all UGS requests are passed to the CAC and then rtps and nrtps requests are processed. For rtps and nrtps only the minimum required slots are guaranteed. Here we present CAC only for DSA. There are minor changes for DSC and DSD requests but the framework remains the same. Figure 5.6 shows our scheduling tree. It also displays the satisfied downlink/uplink slots just below the request. Figure 5.7 shows which requests are admitted when our proposed scheme is applied and when the scheme mentioned in the standard is applied. It is clear from the figure that our proposed scheme admits 11 requests out of 12 whereas the other admits only 8.

Figure 5.6: A scheduling tree with more num-

Figure 5.7: Admitted requests

ber of connections, slots = 20

19

Figure 5.8: Slot allocation for tree shown in Figure 5.6, slots = 20

Figure 5.8 represents the slot allocation at different levels when our proposed scheduling is used with 20 slots per frame. From the above figures it is clear that we are not assigning any slots when there is no transmission. For example, in request 6, we reserved the required slots only at hop 2 and hop 3 due to which we are able to admit more number of requests.

20

1: if Source = 0 then 2:

// Only Downlink

3:

if Slots available at all hops from MBS to destination then Admit this request and update slot information at respective hop

4: 5:

end if

6: else 7:

if Destination = 0 then

8:

// Only Uplink

9:

if Slots available at all hops from source to MBS then Admit this request and update slot information at respective hop

10: 11: 12:

end if else

13:

Find common parent between source and destination node, call it as commonParent

14:

if Source = commonParent then

15:

// Only Downlink

16:

if Slots available at all hops from commonParent to destination then

17:

Admit this request and update slot information at respective hop end if

18: 19:

else if Destination = commonParent then

20: 21:

// Only Uplink

22:

if Slots available at all hops from source to commonParent then Admit this request and update slot information at respective hop

23:

end if

24:

else

25: 26:

// Both uplink and downlink

27:

if Slots available at all hops from source to commonParent AND commonParent to destination then Admit this request and update slot information at respective hop

28:

end if

29:

end if

30: 31: 32:

end if end if

33: end if

Figure 5.9: Call admission control algorithm 21

Chapter 6 Simulation Results Routing As explained in Chapter 4, we started with a single node in the network, the MBS. Then one by one we added a new nodes in the network. A new node can be at any hop. For this we generated a uniform random number between 1 and (maxHops + 1) say h. Then all nodes at hop h, h + 1, h − 1 will send a response to the newly arrived node.

Figure 6.1: Adding new node in a routing tree

Figure 6.1 shows one of the scenario where node 13 is the newly arrived node at hop 4. In the figure, solid lines indicate the link established between two nodes whereas dotted lines indicate the response (i.e. the two nodes have a wireless link in between them). From all the wireless 22

links available, after running parent selection algorithm, node 13 chooses node 3 as its parent (see solid line between node 3 and node 13). Nodes 4, 3 and 8 are at a minimum hop among which node 3 has a higher RSSI value. The routing tree, generated by above routing implementation is used for scheduling is as shown in Figure 6.3. It contains 25 SSs and the MBS at its root.

Figure 6.2: Routing table

Figure 6.3: Routing tree generated and used in scheduling

An efficient scheduling Figure 6.4, 6.5, and 6.6 show the admitted connections as proposed in our scheme. We compared it with the mechanism explained in the IEEE 802.16d standard. From the figure it is clear that our proposed scheme is superior and admits two times more connections than the other. The X-axis indicates the number of connections requested per unit time (λ) whereas Y-axis indicates the acceptance ratio in percentage (%). We assumed that there are around 200 slots per frame to be used for uplink and downlink. We used Poisson distribution for generating connection requests and exponential distribution for deciding connection life time. We plotted the graph for various values of λ. Low value of λ indicates less number of requests per unit time and hence the acceptance ratio is high. As the number of request per unit time increase i.e. as λ increases, the acceptance ratio decreases since we have limited number of slots per frame to admit connection requests.

23

Figure 6.4:

Admitted connections with

Figure 6.5:

slots/request = 5

Admitted connections with

slots/request = 10

Figure 6.6: Admitted connections with slots/request = 25

Also, we can see that as we increase the number of slots per request, number of connections admitted decreases. This is trivial since we have fixed slots per frame.

24

Chapter 7 Proposed Plan We have completed the design and analysis of routing and centralized scheduling for maximizing the overall system throughput and admitting more number of connections. Implementation of routing is also completed. Right now we have used RSSI as the second level parent selection criterion. After some simulation results we will come to know how much weight needs to be assigned for all criteria at second level parent selection. Simulation for admitting more number of connections is also finished. Soon we will present the result for parallel transmission. Various tasks to be completed in MTP stage III are listed below. 1. Assigning weights to the second level parent selection criteria. 2. Simulation to show that the proposed 3-hop parallelism increases the overall system throughput. 3. Call Admission Control with delay guarantees and its simulation.

25

Chapter 8 Conclusion and Future Work In this report we presented a routing and an efficient centralized scheduling scheme for IEEE 802.16 mesh networks. The centralized scheduling algorithm proposed not only increases the overall system throughput by parallel transmissions but also admits more number of connections. The 3-hop parallelism proposed drastically reduces the number of frames required to transmit packets at the last hop. We also show how reserving slots only at the required hop allows us to admit more number of connections. Simulation results show that the number of connections admitted by our scheme is almost twice that of the scheme proposed in the standard Our future work includes simulating this proposed scheduling algorithm and designing call admission control to provide certain QoS guarantees such as delay for real time applications.

26

Bibliography [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] M. Cao, V. Raghunathan, and P. R. Kumar, “A tractable algorithm for fair and efficient uplink scheduling of multi-hop wimax mesh networks”, In Proceedings of 2nd IEEE Workshop on Wireless Mesh Networks (WiMesh 2006), September 2006. [3] Chandra and A. Sahoo, “An Efficient Call Admission Control for IEEE 802.16 Networks”, Technical Report, Available: http://www.it.iitb.ac.in/research/techreport/reports/30.pdf [4] Seungjoon Lee, Girija Narlikar, Martin Pal, Gordon Wilfong, Lisa Zhang., “Admission Control for Multihop Wireless Backhaul Networks with QoS Support”, IEEE WCNC 2006, Las Vegas NV, April 2006. [5] R. Draves, J. Padhye, and B. Zill, “Comparison of Routing Metrics for Static Multi-Hop Wireless Networks”, in ACM SIGCOMM’04, Portland, Oregon, USA, August 2004. [6] H. Wei, S. Ganguly, A. Izmailov, and Z. Haas, “Interference-Aware IEEE 802.16 WiMax Mesh Networks”, in Vehicular Technology Conference, 2005. VTC 2005-Spring. 2005 IEEE 61st , Vol. 5; 3102-3106. [7] F. Jin, A. Arora, J. Hwang, and H.-A. Choi, “Routing and Packet Scheduling for Throughput Maximization in IEEE 802.16 Mesh Networks”, submitted for publication. [8] J. Tao, F. Liu, Z. Zeng, and Z. Lin, “Throughput Enhancement in WiMax Mesh Networks Using Concurrent Transmission”, in International Conference on Wireless Communications, Networking and Mobile Computing, Sep 2005, pp. 871-874.

27

[9] H. Shetiya and V. Sharma, “Algorithms for Routing and Centralized Scheduling to Provide QoS in IEEE 802.16 Mesh Networks”, in WMuNeP ’05: Proceedings of the 1st ACM workshop on Wireless multimedia networking and performance modeling, New York, NY, USA, 2005, pp. 140-149, ACM Press. [10] IEEE P802.16-REVd/D5-2004. “Air Interface for Fixed Broadband Wireless Access Systems”. [11] I. F. Akyildiz, X. Wang, and W. Wang, “Wireless Mesh Network: A Survey”, January 2005. [12] Y. Yuan, H. Yang, S. Wong, S. Lu, W. Arbaugh, “ROMER: Resilient Opportunistic Mesh Routing for Wireless Mesh Networks”, in Proceeding of IEEE Workshop on Wireless Mesh Networks (WiMesh), 2005. [13] R. Draves, J. Padhye, and B. Zill, “Routing in Multi-radio, Multihop Wireless Mesh Networks”, in ACM MobiCom, Philadelphia, PA, September 2004. [14] Wang H., Li W. and Agrawal D.P. “Dynamic admission control and QoS for 802.16 wireless MAN”, Wireless Telecommunications Symposium, 2005, vol. no.pp. 60-66, 6-7 April 2005. [15] Received Signal Strength Indication, Available: http://en.wikipedia.org/wiki/RSSI

28

Suggest Documents