Performance Analysis of the IEEE 802.16 Wireless Metropolitan Area

Performance Analysis of the IEEE 802.16 Wireless Metropolitan Area Network Dong-Hoon Cho, Jung-Hoon Song, Min-Su Kim, and Ki-Jun Han* (* Correspondent author) Department of Computer Engineering, Kyungpook National University, Korea {firecar, pimpo, kiunsen}@netopia.knu.ac.kr, kjhan@bh.knu.ac.kr Abstract In this paper, we propose a new QoS architecture for the IEEE802.16a MAC protocol and present a bandwidth allocation and admission control policy for the architecture. Our architecture provides QoS support to real-time traffic with high priority while maintaining throughput performance to an acceptable level for low priority traffic. Analytical and simulation results assure advantages of our architecture. 1. Introduction The emerging 802.16e and 802.20 standards will both specify new mobile air interfaces for wireless broadband. On the surface the two standards seem very similar, but there are some important differences between them. For one, 802.16e will add mobility in the 2 to 6 GHz licensed bands, while 802.20 aims for operation in licensed bands below 3.5GHz. More importantly, the 802.16e specification will be based on an existing standard (802.16a), while 802.20 is starting from scratch. This means that products based on 802.16e will likely hit the market well before 802.20 solutions. The IEEE approved the 802.16e standards effort in February with the avowed intent of increasing the use of broadband wireless access (BWA) by taking advantage of the "inherent mobility of wireless media." The amendment to 802.16, which is also called the wireless metropolitan area network (WMAN) standard, will enable a single base station to support both fixed and mobile BWA. It aims to fill the gap between high data rate wireless local area networks (WLAN) and high mobility cellular wide area networks (WAN). Broadband Wireless Access (BWA) systems, e.g. IEEE 802.16 standard, provide fixed-wireless access between the subscriber station (residential or business customers) and the Internet service provider (ISP) through the base station. BWA systems complement existing last mile wired networks such as cable modem and xDSL. Due to the upcoming air interface technologies, which promise to deliver high transmission data rates, BWA systems become an attractive alternative. Their main advantage is their fast deployment, which can result in cost savings. For example, such installations can be beneficial in (1) very crowded geographical areas such as cities or in (2) rural areas where there is no wired infrastructure. Without new cable wiring for the whole city, the antenna of the base station and subscriber customers are easily set up at the rooftop of their buildings to form the wireless network. BWA systems are expected to support quality of service (QoS) for real time applications such as video conferencing, video streaming, and voice over IP. Such applications are delay and delay variation sensitive. In other words, the quality of the application is severely degraded because the packets suffer large delays and delay variation.[1] IEEE 802.16 media access control, which is based on the concepts of connections and service flows, specifies QoS signaling mechanisms (per connection or per station) such as bandwidth requests and bandwidth allocation. [1] However, IEEE 802.16 standard left the QoS based packet-scheduling algorithms that determine the uplink and downlink bandwidth allocation, undefined. This paper proposes an efficient QoS architecture, based on priority scheduling and dynamic bandwidth allocation. The system performance is analytically evaluated and is verified through a simulation. The remaining of this paper is organized as follows. Section 2 reviews the BWA Systems and IEEE 802.16 MAC Protocol. In section 3, we describe the existing IEEE 802.16 QoS architecture. We present a new QoS architecture for QoS support to real-time traffic with high priority while maintaining throughput performance to an acceptable level for low priority traffic in section 4. Section 5 provides simulation results of our QoS architecture and we conclude in Section 6. Proceedings of the First International Conference on Distributed Frameworks for Multimedia Applications (DFMA’05) 0-7695-2273-4/05 $ 20.00 IEEE 2. BWA Systems and IEEE 802.16 MAC Protocol IEEE 802.16 architecture consists of two kinds of fixed (non-mobile) stations: subscriber stations (SS) and a base station (BS). The BS regulates all the communication in the network, i.e. there is no peer-to- peer communication directly between the SSs. Each SS can deliver voice and data using common interfaces, such as plain and telephony service, Ethernet, video, VoD and other services with different QoS requirements. [2] Fig. 1. Broadband wireless access system architecture The communication path between SS and BS has two directions: uplink channel (from SS to BS) and downlink channel (from BS to SS). The downlink channel, defined as a direction of data flow from the BS to the SSs, is a broadcast channel, while the uplink channel is a shared by SSs. Time in the uplink channel is usually slotted (mini-slots) called by time- division multiple access (TDMA), whereas on the downlink channel BS uses a continuous time-division multiplexing (TDM) scheme as shown in Fig. 2. [2] Fig. 2. IEEE 802.16 TDD frame structure The BS dynamically determines the duration of these subframes. Each subframe consists of a number of time slots. SSs and BS have to be synchronized and transmit data into predetermined time slots. The uplink channel is divided into a sequence of mini-slots. Since all the SSs have synchronized with the BS clock, the BS controller can uniquely number each mini-slot that will arrive at a particular SS. SSs send requests in the uplink channel to BS. In the downlink channel, the BS uses a combination of acknowledgem (ACK) and grant (GR) mini-slots to acknowledge requests from SSs and to grant access to data slots. Here, we will specify some basic characteristics of the common IEEE 802.16 MAC protocol to create a framework for designing the QoS architecture. Note that only MAC protocols that are necessary for the design of the QoS architecture are described. To support QoS, IEEE 802.16 defines four QoS services: Unsolicited Grant Service (UGS); Real-Time Polling Service (rtPS); Non-Real-Time Polling Service (nrtPS) and Best Effort (BE) service. UGS service is prohibited from using any contention requests, there is no explicit bandwidth requests issued by SS. The BS must provide fixed size data grants at periodic intervals to the UGS flows. UGS service can be used for constant bit-rate (CBR) for CBR-like service flows such as T1/E1. However, the reserved bandwidth may be wasted when a corresponding UGS flows is inactive. The rtPS and nrtPS flows are polled through the unicast request polling. However, the nrtPS flows receive few request polling opportunities during network congestion and are allowed to use contention requests, while the rtPS flows are polled regardless of network load and frequently enough to meet the delay requirements of the service flows, moreover, rtPS flows prohibited from using any contention requests. Real-time Polling Services (rtPS) can be used for rt- VBR-like service flows such as MPEG video, Non- real-time Polling Service (nrtPS) can be used for not- real-time service flows with better than best effort service such as bandwidth-intensive file transfer [2]. 3. QoS Architecture for IEEE 802.16 MAC protocol IEEE 802.16 can support multiple communication services (data, voice, video) with different QoS requirements. The media access control (MAC) layer defines QoS signaling mechanisms and functions that can control BS and SS data transmissions. On the downlink (from BS to SS), the transmission is relatively simple because the BS is the only one that transmits during the downlink subframe. The data packets are broadcasted to all SSs and an SS only picks up the packets destined to it. One of the modes of uplink arbitration (from SS to BS) uses a TDMA MAC. The BS determines the number of time slots that each SS will be allowed to transmit in an uplink subframe. This information is broadcasted by the BS through the uplink map message (UL-MAP) at the beginning of each frame. UL-MAP contains information element (IE), which include the transmission opportunities, i.e. the time Proceedings of the First International Conference on Distributed Frameworks for Multimedia Applications (DFMA’05) 0-7695-2273-4/05 $ 20.00 IEEE slots in which the SS can transmit during the uplink subframe. After receiving the UL-MAP, each SS will transmit data in the predefined time slots as indicated in IE. The BS uplink-scheduling module determines the IEs using bandwidth request PDU (BW-request) sent from SSs to BS. [1] In IEEE 802.16 standard, there are two modes of transmitting the BW-Request: contention mode and contention-free mode (polling). In contention mode, SSs send BW-Request during the contention period. Contention is resolved using back-off resolution. In contention-free mode, BS polls each SS and SSs reply by sending BW-request. Due to the predictable signaling delay of the polling scheme, contention-free mode is suitable for real time applications. IEEE 802.16 defines the required QoS signaling mechanisms described above such as BW-Request and UL-MAP, but it does not define the Uplink Scheduler, i.e. the mechanism that determines the IEs in the UL-MAP. Fig. 3 shows the existing QoS architecture of IEEE 802.16. Uplink Bandwidth Allocation scheduling resides in the BS to control all the uplink packet transmissions. Since IEEE 802.16 MAC protocol is connection oriented, the application first establishes the connection with the BS as well as the associated service flow (UGS, rtPS, nrtPS or BE). BS will assign the connection with a unique connection ID (CID). The connection can represent either an individual application or a group of applications such as multiple tenants in an apartment building (all in one SS) sending data with the same CID. [1] IEEE 802.16 defines the connection signaling (connection request, response) between SS and BS but it does not define the admission control process. All packets from the application layer in the SS are classified by the connection classifier based on CID and are forwarded to the appropriate queue. At the SS, the Scheduler will retrieve the packets from the queues and transmit them to the network in the appropriate time slots as defined by the UL-MAP sent by the BS. The UL-MAP is determined by the Uplink Bandwidth Allocation Scheduling module based on the BW- request messages that report the current queue size of each connection in SS [1] . In summary, IEEE 802.16 defines: (1) the signaling mechanism for information exchange between BS and SS such as the connection set-up, BW-request, and UL-MAP and (2) the uplink scheduling for UGS service flow. IEEE 802.16 does not define: (1) the uplink scheduling for rtPS, nrtPS, BE service flow and (2) the admission control. Fig. 3. QoS architecture of IEEE 802.16 4. A new QoS Architecture of IEEE 802.16 Now, we propose a QoS architecture that completes the missing parts in the IEEE 802.16 QoS architecture. As shown in Fig 4, at the BS we add a detailed description of the Uplink Bandwidth Allocation Scheduling part (scheduling algorithm that which supports all types of service flows), and admission control part. At the SS we add a traffic management module. For each of the UGS, rtPS, nrtPS, BE service, multiple connections are aggregated into their respective service flow. The schedule process is divided into two steps. The first step is performed at the BS according to the information of the request from the SS. Then the uplink scheduler of SS is responsible for selection of appropriate packets from all queues and sends them through the uplink data slots granted by the Packet Allocation Module of BS. The BS must provide fixed size data grants at periodic intervals to the UGS flows Fig. 4. Proposed QoS architecture of IEEE 802.16 Here is a brief description of the connection establishment using the QoS architecture in Fig 4: (1) An application that originates at an SS establishes the connection with BS using Proceedings of the First International Conference on Distributed Frameworks for Multimedia Applications (DFMA’05) 0-7695-2273-4/05 $ 20.00 IEEE connection signaling. The application includes in the connection request the traffic contract (bandwidth and delay requirement). (2) The admission control part at the BS accepts or rejects the new connection. (3) If the admission control part accepts the new connection, it will notify the Uplink Bandwidth Allocation Scheduling part at the BS and provide the token bucket parameters to the traffic management module at the SS. After the connection is established, the following steps are taken: (1) Traffic management enforces traffic based on the traffic contract of the connection. (2) At the beginning of each time frame, the data packet analysis module collects the queue size information from the BW-requests received during the previous time frame. The data packet analysis module will process the queue size information and update the traffic management table. (3) The packet allocation module retrieves the information from the traffic management module and generates the UL-MAP. (4) BS broadcasts the UL-MAP to all SSs in the downlink subframe. (5) The scheduler of SS transmits packets according to the UL-MAP received from the BS. 4.1 Uplink Scheduler of SS The uplink scheduler of SS transmits data PDU using uplink data slot granted by BS. For rtPS connection, SS scheduler transmits UGS packets to append virtual packet arrival time of rtPS. Data Packet Analysis Module (DPAM) of BS received uplink data from SS obtains virtual packet arrival time of rtPS at UGS packets of each SS. DPAM can find the rtPS deadline information. Based on this deadline information, the Uplink Bandwidth Allocation Scheduling of BS will know exactly when to schedule packets such that packet delay requirements are met. We apply the arrival-service curve concept to determine the packet arrival and deadline. The packet deadline is their arrival time plus the maximum delay requirement of the connection. Therefore, BS can determine order of Poll based on this information. The nrtPS uses either contention-free mode or contention mode. If SS receives poll by BS and does not have rtPS connection in queue, the nrtPS connection can transmit BW-request. The BE uses only contention mode. If SS wins contention mode, SS can transmit BW-request for BE connection. In IEEE 802.16, BS can control contention mini-slot size in frame. If contention mini-slot size increases, uplink data slot will decrease. If contention mini-slot size decreases, uplink data slot will increase. Therefore, we propose that BS determine a efficient contention mini- slot size. Contention resolution proposed in IEEE 802.16 is similar to general contention resolution of wireless network. The mandatory method of contention resolution that shall be supported is based on a truncated binary exponential backoff, with the initial backoff window and the maximum backoff window controlled by the BS. 4.2 Uplink Bandwidth Allocation Scheduling After SSs transmit UGS packets by uplink data slot, Data Packet Analysis Module (DPAM) of BS separates UGS data and virtual packets arrival time of rtPS. This information manages at Traffic Management Module and uses the Polling schedule in next frame. The rtPS is time-bounded data. Therefore, BS is apt to give Poll to SS coming up close to deadline. In this work, we assume BS has the ability to detect collision in each contention period mini-slot. The BS broadcasts a common back-off window size “B” to all the competing SSs. SSs will then randomly choose a reservation slot numbered between 1 and B to transmit its request. We assume that there are N SSs in the system and BS broadcasts a back-off window size B. Since each user will choose between 1st and Bth reservation slots to send its bandwidth reservation, the probability of choosing a given slot is p=1/B. As a result, the probability of a given slot that is not selected by any SS is given by: (1) The probability of a successful transmission is equal to the probability that a single user selects a given slot. Thus, the system throughput is given by: (2) To maximize system throughput, we have to get: (3) Proceedings of the First International Conference on Distributed Frameworks for Multimedia Applications (DFMA’05) 0-7695-2273-4/05 $ 20.00 IEEE In other words, the maximum throughput can be obtained when BS broadcasts a back-off window size (B) which is equal to the number of competing SSs (N). 4.3 Analytical model for channel utilization Here, we find out an analytical model for channel utilization. We assume that there are k classes of priority queues:G Class 1 is the highest priority traffic, and class 2 is the second highest priority traffic, and class k means the lowest priority. We also assume that arrival events are mutually independent. Let C and denote the server capacity and channel utilization for each class i, respectively. Then we have (4) In our scheme, the higher priority class is allocated the bandwidth first, and then the lower priority class is allocated the remaining bandwidth late. For example, the capacity of class 2 uses the remainder of capacity left over class 1. Similarly, the server allocates the remainder of capacity to class 4 after class 1, 2 3 are allocated. So, we have (5a) (5b) (5c) (5d) where kO is offered load for class k and ][ kE W is the average service time for class k. Using these equations, we can get channel utilization for each class of priority traffic 5. Simulations In this section, we evaluate performance of our scheme for IEEE 802.16. The system model for analysis consists of one base station and numbers of subscriber stations (SS). Several assumptions have been made to reduce the complexity of the simulation model: ˍ The effects of propagation delay are neglected. ˍ The channel is error-free that means that each transmitted packet was successfully and correctly received at its destination. ˍ The BS has the ability to detect collision. In addition, each SS is assumed to be a Poisson traffic source and the packet size (including overhead) is variable. The parameters used for performance evaluation are listed in Table 1. TABLE 1. SIMULATION PARMETER Figure 5 shows channel utilization obtained by simulation experiments and the analytical model given by Eq. (5a)~(5d). Fig. 5(a) shows channel utilization when we assume that there is no limit of the bandwidth that the highest priority traffic can take. On the other hand, Figure 5(b) shows channel utilizations when a fixed quota is allowed for the UGS flow and the remaining bandwidth is used for the other three flows. We can see that the UGS flows do not increase any more above some value because the BS provides fixed size data grants to the UGS flows at periodic intervals. Fig. 5(c)~5(f) compares analytical and simulation results of channel utilization for four different priorities of packets. This figure indicates that our analytical model is simple, nevertheless accurate. The channel utilization of the high priority traffic increases linearly because it is not affected by the transmission of lower priority traffic. ͡ ͟͢͡ ͣ͟͡ ͤ͟͡ ͥ͟͡ ͦ͟͡ ͧ͟͡ ͨ͟͡ ͩ͟͡ ͟͡͡ ͢ ͣ͟͡ ͥ͟͡ ͧ͟͡ ͩ͟͡ ͢