A VARIABLE BIT RATE RESOURCE ALLOCATION ALGORITHM FOR WIRELESS ATM Jean-Franc¸ois Frigon, Henry C. B. Chan and Victor C.M. Leung Department of Electrical and Computer Engineering, The University of British Columbia, Vancouver, B.C., Canada, V6T 1Z4 E-Mail: fjeanf, chunc,
[email protected] Abstract A resource allocation algorithm for variable bit rate (VBR) sources using the Dynamic Reservation TDMA (DRTDMA) medium access control protocol is proposed in this paper. DR-TDMA combines the advantages of distributed access and centralized scheduling to multiplex VBR traffic efficiently over a wireless ATM channel, and can be easily extended to integrate constant bit rate, VBR, and available bit rate traffic. We present a novel rate based scheduling algorithm which maintains in the base station a virtual status of the access queue in each mobile to control channel access by the mobile. A unique cell control algorithm is incorporated to provide guaranteed or best effort service to each VBR source depending on whether the current cells conform to the traffic contract. Simulation results show that DR-TDMA can achieve a high throughput in the range of 93 to 98% while maintaining a reasonable quality of service. 1. Introduction In recent years, we have seen a rapid proliferation of high performance portable computers and an increasing popularity of multimedia applications. We can therefore expect an emerging demand to connect these devices to the fixed network to transfer voice, video, data and multimedia traffic. The wireless network will be required to be compatible with the wired broadband communication network. Asynchronous Transfer Mode (ATM) was recommended by ITUT to be the transfer protocol of the emerging Broadband Integrated Services Digital Network. The concept of wireless ATM (WATM) was introduced in 1994 to extend the capabilities of ATM over the wireless channel [1]. A major issue related to the realization of WATM networks is the selection of a Medium Access Control (MAC) protocol that will efficiently and equitably allocate the scarce and valuable radio medium among the competing mobile nodes while respecting the Quality-of-Service (QoS) requirements of each admitted connection. For compatibility with the wired ATM network, the WATM MAC protocol must support the standard ATM service classes for Constant Bit Rate (CBR), Variable Bit Rate (VBR) and Avail This work was supported by a postgraduate scholarship and grant OGP0044286 from the Natural Sciences and Engineering Research Council of Canada.
able Bit Rate (ABR) traffic. In the past couple of years, several projects have developed MAC protocols for wireless ATM networks [2, 3, 4, 5] and most of these protocols are designed to support the ATM multimedia traffic. In [6], we proposed the Dynamic Reservation TDMA (DR-TDMA) MAC protocol that is capable of maintaining a high channel utilization while respecting the QoS requirements of each connection depending on its ATM traffic class. DR-TDMA incorporates the framed pseudo-Bayesian priority (FPBP) Aloha algorithm [7] to manage the control slot access. We also introduced resource allocation algorithms for CBR, voice and data traffic. To allow DR-TDMA to efficiently support multimedia applications, we propose in this paper a novel predictive approach for VBR resource allocation where the base station estimates from a limited number of parameters the current resource requirements of the VBR connections. A similar predictive approach has been proposed in [4]. However, we are able to significantly increase the performance with the method presented in this paper, by forcing the VBR source to be rate-controlled. Our method also employs a unique cell control algorithm that complies with the ATM Forum specifications [8]; such compliance has not been accomplished before in other WATM proposals. This cell control algorithm enforces the conformance of the VBR flow with the connection’s traffic parameters (sustained cell rate, peak cell rate and burst tolerance) for the purpose of providing QoS guarantees. The paper is organized as follows. Section 2 gives an overview of the DR-TDMA protocol. In Section 3, the VBR source model is presented. The VBR resource allocation algorithm is described in Section 4. A performance evaluation of DR-TDMA with VBR traffic by simulations is presented in Section 5. Finally, Section 6 concludes the paper. 2. Dynamic Reservation TDMA Protocol A dynamic reservation TDMA protocol is adopted to multiplex multimedia ATM connections over the time division duplex (TDD) radio channel [6, 9]. Figure 1 illustrates the frame structure employed by the DR-TDMA protocol. The fixed length DR-TDMA frame is time-duplexed into an uplink and downlink channel and the boundary between these two parts is dynamically adjusted as a function of the traffic load. The downlink and uplink channels are dynamically divided into control and data transmission periods. Slots as-
Fixed Length Frame Uplink
Downlink
... Modem Frame Downlink control Data slots preamble header mini-slots
... Uplink control mini-slots
Data slots
Figure 1: DR-TDMA frame structure
signed for control purposes are further divided into control mini-slots used to transmit control information. System parameters adopted for the DR-TDMA protocol are according to the literature [1, 4] and are summarized in Table 1. Table 1: DR-TDMA MAC frame parameters Channel bit rate 8.528 Mbps Frame duration 2 ms Uplink data slot size 60 bytes Uplink control slot size 20 bytes Preamble size 16 bytes Frame header 16 bytes Number of slots per frame 35 Number of control mini-slots per slot 3
In the uplink channel, control slots provide a mechanism for a mobile station to send a reservation request during the contention phase of the connection, while the data slots supply it with contention-free bandwidth resources during the data transmission phase. Uplink control packets are sent in contention-mode according to the FPBP protocol described in [6] and they provide the base station with the traffic characteristics and source status of the corresponding mobile stations’ connections. After completing the contention procedure, the mobile station can use the data slots assigned by the base station without undergoing further contentions. While a mobile station has access to reserved data slots, traffic parameters and status information can also be sent to the base station piggybacking on the data packets. The base station will use these traffic parameters to allocate uplink data slots to each reserving station according to the resource allocation algorithms. The base station also determines according to the estimated contention traffic, the number of slots to be allocated for control purpose [6]. When a connection has successfully sent its request, it enters the data transmission phase and monitors downlink control slots in each subsequent frames to receive its slot assignments. 3. VBR Source Model The VBR source model presented in this section is useful not only for the performance analysis of the DR-TDMA protocol, but it also forms the basis of the VBR resource allocation algorithm proposed in Section 4. To enable the implementation of an efficient VBR allocation algorithm which meets the QoS specifications, traffic
shaping as defined by the ATM Forum is mandatory for the VBR sources. The cell control algorithm is similar to what would be required for a VBR source connected to an ATM network and the source rate control is in conformance with the frame structure of a typical video coder such as MPEG. A VBR source generates a traffic flow at a rate that varies with time. If the transmission rate varies continuously in an uncontrolled manner, it would require much control traffic from the mobile station to keep the base station updated about the buffer state in the VBR source. Fortunately, real VBR sources such as MPEG video encoders do not generate traffic flows which continuously vary with time. Instead, their bit-rates typically vary among a fixed number of possible rates [10]. Also, a general VBR traffic flow can be rate-controlled using algorithms as presented in [11]. We therefore consider a discrete rate VBR model that gives the basis for efficient slot scheduling with a minimal amount of control traffic. In this model, the source traffic flow when cell c arrives has an instantaneous cell rate Rc which is an integer multiple l c = 0; 1; : : : of the basic source cell rate Rb , i.e., Rc = l c Rb . After the arrival of cell c, the cell arrival rate changes to the value Rc+1 = l c+1 Rb (note that Rc+1 can be equal to Rc ) and packet c + 1 will arrive 1=(l c+1 Rb ) seconds after cell c’s arrival. The above discrete rate VBR model can be represented mathematically by a superposition of S ON/OFF sources. Each ON/OFF source s (s = 1; : : : ; S) alternates between the ON and OFF states. Durations of ON and OFF states are assumed to be exponentially distributed with means tbvbr and tivbr , respectively. Although this mathematical model is not representative of every types of VBR sources, it provides a easy means to evaluate the efficiency of the proposed protocol and demonstrate its properties. The parameter l c is equal to the number of sources simultaneously in the ON state at discrete time c. When a VBR connection is admitted by the WATM network, it is required to specify the parameters of its traffic flow. Cells that conform to the specified traffic contract receive a high QoS while non-conforming cells are served on a best-effort basis without any QoS guarantee. Nonconforming cells are not necessarily discarded in order to take advantage of unused resources in the network. When a packet (cell) arrives from the VBR rate-controlled source, the cell control algorithm shown in Figure 2 determines if the packet conforms with the VBR traffic parameter. An arriving packet is considered to comply with the specified traffic parameter if, when the packet arrives from the source, there is a token in the guaranteed token pool. In this case, a token is removed from the guaranteed token pool and the packet is tagged as a guaranteed packet and receives priority service which guarantees a high QoS. Otherwise, the packet is tagged as a best-effort packet (QoS is not guar-
Remove one token from the guaranteed token pool and set MAC field to indicate guaranteed packet Yes
Arriving packets from the rate-controlled VBR source
Is there a token in the guaranteed token pool?
No
Set MAC field to indicate best-effort packet
Output buffer
Guaranteed token pool with depth W
Arriving guaranteed tokens at rate gRb (turned away if there is W tokens in the token pool)
Figure 2: Cell Control Algorithm
anteed). Cells are queued in the output buffer until the mobile station receives slot allocations for their transmissions over the wireless channel. Tokens arrive at a constant rate of gRb and are queued in the guaranteed token pool, which can store a maximum of W tokens. Tokens arriving at the guaranteed token pool when it is full are lost. In summary, for the DR-TDMA protocol, a VBR source in the WATM network is described by the following traffic parameters: (1) Rb : the basic cell rate of the source; (2) g: the parameter characterizing the sustained cell rate gRb guaranteed to the source; (3) W : the depth of the guaranteed token pool; (4) Peak Cell Rate (PCR): the maximum allowed cell rate determined by the rate-controlling algorithm; (5) MTD: the maximum cell transfer delay from the best-effort or guaranteed buffer to the base station; i.e., cells queued in the buffers longer than MTD are dropped. 4. Slot Allocation Algorithm for VBR Traffic The rate-based VBR allocation algorithm that we propose maintains in the base station the virtual status of each admitted VBR connection and allocates slots according to its specified QoS and virtual status. The current cell arrival rate of a VBR source is transmitted from the mobile station to the base station either piggybacked to a data packet or explicitly in a control packet and is used by the scheduler to predict the buffer status of the connection (number of packets, types of packets - guaranteed or best effort, arrival times, number of tokens in guaranteed token pool). When a new VBR connection is established, the mobile station sends a request control packet containing the VBR source traffic parameters in the contention period using the FPBP protocol. The request packet also contains the initial status information of the connection (cell rate, number of tokens in the bucket, : : : ). A MAC field in each data packet is used to indicate the packet type and the change in source rate-level after the packet has left the VBR source. The overhead required for the allocation algorithm in data packets is thus relatively small. One bit can indicate the packet type while the rate
change can be encoded in a small number of bits. The base station can predict, using the information about the current transmission rate of the VBR source, when a data packet arrival occurs at the mobile and its type. These predictions are done by reproducing the source arrival process in the base station scheduler. If a rate change occurs for one of the predicted packets, then subsequent predictions will be erroneous. However, when the base station subsequently receives the packet with the rate change indication, this information as well as information stored with each virtual packet (current arrival rate, arrival time, : : : ) allows the virtual buffer status prediction to be updated correctly starting from this packet. When the cell arrival rate is equal to zero, or a packet with a non-zero rate change field is dropped in the mobile station due to its MTD being exceeded, the base station can no longer correctly predict cell arrivals and buffer status. When such a situation is detected, the status of the first packet of the next burst or the lost rate change information must be transmitted in a control packet. Furthermore, to avoid any error in the prediction algorithm, if a data packet with a nonzero rate change is waiting for transmission while transmission of a control packet due to the above reasons is pending, then the rate change information associated with the data packet should also be sent in the control packet. When the base station receives a request in a control packet, it must retrieve the information stored with the corresponding virtual packet in order to predict the new virtual buffer status. We therefore need to number packets. To preserve packet sequencing, packets are numbered according to their order of arrival from the source. Since the MAC layer preserves sequencing of all transmitted packets, except that packets could be lost due to MTD being exceeded, packet numbers needs to be transmitted only in control packets and are not included in data packets. We also need to store virtual status information about transmitted packets, including lost packets, in a “transmitted” queue in the base station for later packet information retrieval as required by the prediction algorithm. When packet arrival resumes after the cell arrival rate has been zero, the base station has no virtual packet corresponding to the first packet in the burst. The control packet must therefore contain the packet number of the first arrival, the packet type, the packet arrival time, the number of tokens in the guaranteed token pool and the packet number of the last packet transmitted in the current frame. All other control packets that contain information about data packets for which the previous rate was non-null, include the packet numbers of up to three data packets with non-zero ratechange values, the corresponding rate-change values, and the packet number of the last packet transmitted in the current frame. Prediction updates in the base station are done
5. Performance Evaluation To focus our evaluation on the efficiency of the uplink MAC protocol, we assume a radio channel without errors and fading and that downlink control slots do not consume any transmission bandwidth. The throughput is defined as the ratio of the average number of slots used for data packet transmissions per frame (excluding control packets) to the total number of slots available per frame. The offered load and achieved throughput thus include cell headers but do not include control packets. In this section we only present a subset of the simulation results that have been obtained (see [9] for additional results). We have simulated the DR-TDMA protocol with
−1
10
g=5 − Total loss g=5 − Gua. loss g=10 − Total. loss g=10 − Gua. loss g=25
−2
VBR cell loss rate
10
−3
10
−4
10
7
7.5
8
8.5
9
9.5
10
10.5
11
Number of connections
Figure 3: Cell loss rate as a function of the number of VBR connections. 45
g=5 − Gua. delay g=10 − Gua. delay g=25
40
35
30
Delay (ms)
as explained above except that we need the number of the last received packet in order to know if a predicted packet should be put in the virtual buffers for scheduling or in the transmitted packet queue. For each packet received at the base station, depending on its type, the next guaranteed or best-effort packet is removed from the virtual buffer and placed in the transmitted packet queue. At the end of each frame, the base station updates the virtual status prediction of each VBR connection based on the information received during the frame either piggybacked to a data packet or directly in a control packet. Furthermore, packets for which the MTD is exceeded are removed from the virtual queue at the base station and from the actual transmission queue at the mobile station. When the connections’ virtual status has been updated in the base station, the scheduler allocates slots to VBR connections according to the virtual buffer status and connections QoS. The slot allocation algorithm is divided into two parts: guaranteed allocation and best-effort allocation. For the guaranteed service, VBR connections with guaranteed packets in their virtual queue are sorted in a list in increasing order of TTE (time to expiry) where TTE = MTD (current time arrival time of the first guaranteed packet in the queue). The scheduler allocates, accordingly to this list, the S available slots to the VBR connections with waiting guaranteed packets. Then, if there are still available slots, the scheduler allocates the remaining slots to the VBR connections with best-effort packets in their virtual queue according to a fair bandwidth allocation algorithm [9]. When the allocations for guaranteed and best-effort packets have been made, the base station announces in the downlink control slots the number of slots allocated to each mobile stations. Each VBR connection will transmit the packets in the same order as they arrived in its buffer, with priority given to guaranteed packets. If a best-effort packets can not be transmitted in the current frame and a guaranteed packet with a higher number is transmitted in the current frame then the best-effort packet is dropped.
25
20
15
10
5
0
1
2
3
4
5
6
7
8
9
10
11
Number of connections
Figure 4: Cell delay as a function of the number of VBR connections.
VBR connections only to evaluate the performance of the novel VBR resource allocation algorithm that we have presented in this paper. 667 Kbps VBR connections with a basic source interarrival time (1=Rb) of 3.6 ms, 25 ON/OFF sources with an average ON state length of 100 ms and an average OFF state length of 300 ms were used for these simulations. The token pool depth W was set to 1 and the MTD was 50 ms. Figures 3, 4 and 5 respectively show the VBR cell loss rate, delay and throughput as functions of the number of VBR connections for guaranteed parameter g values of 5, 10 and 25. We can see that the total cell loss rate remains low (below 1%) for an offered load below 93%. We observe that when the parameter g is decreased, the total loss rate increases slightly but, as expected, there is a significant improvement of the cell loss rate for guaranteed packets. It should be noted that for g = 5, no guaranteed packet has been lost in the simulations. This explains why in Figure 3 the cell loss rate curve for guaranteed packets with g = 5 is missing. This result confirms that our protocol is able to deliver a better QoS to guaranteed traffic. Another useful performance measure introduced in [4] is the Allocation Efficiency (AE) defined as the fraction of data slots allocated to VBR traffic that are actually used for
call that the goal of the parameter W is to offer more flexibility on the definition of the guaranteed traffic characteristics. The ON-OFF model used to characterized the VBR sources is known to be slowly varying which can explain the small impact of W . For burstier traffic, the flexibility that W offers to accommodate a larger burst length can help to improve QoS by better characterizing the VBR traffic.
1
g=5 g=10 g=25
0.9
0.8
Throughput
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1
2
3
4
5
6
7
8
9
10
11
Number of connections
Figure 5: Throughput as a function of the number of VBR connections.
transmissions. A high AE index indicates a good match between the virtual status at the base station scheduler and the real status of the respective mobile station. Furthermore, a higher AE will allow the system to support more data traffic. For the simulated conditions we have obtained AE values superior to 99%. For comparison purposes, the connection and channel parameters used for the simulations with these VBR connections have been set similar to the one used by Raychaudhuri et. al. to evaluate their MAC protocol in [4]. The results presented in figures 4 and 5 can thus be compared with the results reported in [4]. At low throughput, the DR-TDMA and NEC protocols both show a stable delay around 2 ms. However, if we take a target delay of 5 ms, the maximum throughput of the protocol proposed by Raychaudhuri et. al. is around 64% while we can attain a throughput of 90% with the DR-TDMA protocol. Similarly, if we target a 10 ms delay, the maximum throughput that the NEC protocol can achieve is 72% while our DR-TDMA protocol can reach a throughput in the range of 93% and 98% depending on the value of g. For the AE performance, our DR-TDMA protocol shows a performance superior to 99.1%, while the protocol presented by Raychaudhuri et. al. has an AE varying between 75% and 80%. From these results, we can see that our DR-TDMA MAC protocol outperforms the bandwidth allocation algorithm proposed in [4] for VBR traffic. In the previous simulations, we have used a guaranteed token pool depth W of one token. The parameter W allows a better adjustment of the guaranteed traffic characteristics. We have therefore simulated the DR-TDMA protocol with with token pool depth W of 1 and 10 tokens. With W = 10 the guaranteed traffic is higher, therefore it was observed that the total cell loss rate improves slightly while the cell loss rate of guaranteed packets increases similarly to what was observed previously when g was increased. While the impact of the guaranteed token pool depth value on the cell loss rate is small for these traffic conditions, we should re-
6. Conclusions In this paper we have introduced a new rate-based VBR cell scheduling algorithm as well as a companion cell control algorithm for the DR-TDMA WATM MAC protocol. We have presented simulation results that clearly demonstrate that the protocol can offer a high maximum throughput (in the range of 90-95%) and low delays. In comparison with the performance of previous proposals, our results clearly showed that DR-TDMA performs better than what has previously been reported in the literature. The proposed VBR allocation algorithm can be easily integrated with the CBR and ABR allocation algorithm presented in [6] in order to efficiently integrates multiple BISDN traffic classes over a WATM link. The DR-TDMA WATM MAC protocol can therefore provide a high throughput efficiency while respecting the QoS requirements of the different ATM traffic classes. References [1] D. Raychaudhuri and N. D. Wilson, “ATM-Based transport architecture for multiservices wireless personal communication networks,” IEEE JSAC, vol. 12, pp. 1401–1414, Oct. 1994. [2] O. Kubbar and H. T. Mouftah, “Multiple access control protocols for wireless ATM: Problems definition and design objectives,” IEEE Comm. Mag., vol. 35, pp. 93–99, Nov. 1997. [3] D. Raychaudhuri, L. J. French, R. J. Siracusa, S. K. Biswas, R. Yuan, P. Narasimhan, and C. A. Johnston, “WATMnet: A prototype wireless ATM system for multimedia personal communication,” IEEE JSAC, vol. 15, pp. 83–95, Jan. 1997. [4] S. K. Biswas, D. Reininger, and D. Raychaudhuri, “UPC base bandwidth allocation for VBR video in wireless ATM,” in Proc. ICC’97, (Montr´eal, Canada), June 1997. [5] F. Bauchot et Al., “MASCARA: A MAC protocol for wireless ATM,” in Proc. ACTS Mobile Summit, (Granada, Spain), pp. 17–22, Nov. 1996. [6] J. F. Frigon, H. C. B. Chan, and V. C. M. Leung, “Data and voice integration in DR-TDMA for wireless ATM networks,” in Proc. ICC’99, (Vancouver, Canada), June 1999. [7] J. F. Frigon and V. C. M. Leung, “A pseudo-Bayesian Aloha algorithm with mixed priorities for wireless ATM,” in Proc. of PIMRC’98, (Boston, MA), September 1998. [8] The ATM Forum, “The ATM Forum traffic management specification.” v. 4.0, 1996. [9] J. F. Frigon, “Dynamic reservation TDMA medium access control protocol for wireless ATM networks,” Master’s thesis, University of British Columbia, August 1998. [10] H. Heeke, “A traffic-control algorithm for ATM networks,” IEEE Trans. on Circuits and Systems for Video Technology, vol. 3, pp. 182–189, June 1993. [11] K. Joseph and D. Reininger, “Source traffic smoothing and atm network interfaces for vbr mpeg encoders,” in Proc. of ICC’95, vol. 3, (Seattle, WA), pp. 1761–1767, June 1995.
