UBICC, the Ubiquitous Computing and Communication Journal [ISSN 1992-8424], is an international scientific and educational organization dedicated to advancing the arts, sciences, and applications of information technology. With a world-wide membership, UBICC is a leading resource for computing professionals and students working in the various fields of Information Technology, and for interpreting the impact of information technology on society.
DIFFERENT MAC PROTOCOLS FOR NEXT GENERATION WIRELESS ATM NETWORKS Sami A. El-Dolil Dept. of Electronic and Electrical Comm. Eng., Faculty of Electronic Eng, Menoufya Univ. Msel_dolil@yahoo.com ABSTRACT This paper presents a comparison between three proposed Medium Access Control (MAC) Protocols for next generation multimedia wireless ATM (WATM) networks. To support the ATM CBR, VBR, ABR services to end users, a MAC protocol must be able to provide bandwidth on demand with suitable performance guarantee. The protocols have been proposed to efficiently integrate multiple ATM traffics over the wireless channel while achieving high channel utilization. The objective of the comparison is to highlight the merits and demerits of the three proposed protocols. Keywords: Medium access control protocol and ATM network. 1 INTRODUCTION wireless ATM networks. The three protocols are as follow. Asynchronous transfer mode (ATM) was 1. Dynamic Allocation TDMA MAC Protocol for recommended by the International Wireless ATM Networks. Telecommunication Union (ITU-T) to be the transfer 2. An Intelligent MAC Protocol for next generation protocol of the broadband integrated services digital Wireless ATM Networks. network (B-ISDN). The concept of wireless ATM 3. Contention and Polling based Multiple Access (WATM) was introduced to extend the capabilities of Control with minimum Piggybacking for ATM to wireless arena in . Wireless ATM Network. A major issue of WATM network is the Three performance metrics, namely cell loss selection of a medium access control (MAC) probability, average cell delay, and throughput, are protocol that will efficiently allocate the scarce radio considered. Section II gives an overview and resources among the competing mobile stations description of the proposed protocols. In section III, while satisfying the QoS required for each admitted the source models are identified. Section IV, connection. describes the resource allocation algorithm. An Several MAC protocols are proposed for evaluation of the performance of the proposed wireless ATM network  – . In , a novel protocols is presented in section V. Finally, section predictive approach is used to estimate the current VI concludes the paper. requirements for the connections. The variable bit rate (VBR) traffic is divided into guaranteed and 2 SYSTEM DESCRIPTION best effort traffic while the time to expiry algorithm is adapted for voice and VBR slot allocation. In , 2.1 Air Interface Frame Structure the leaky bucket algorithm with priority as well as The proposed protocols use frequency division the cell train concept achieves a fair and efficient duplex (FDD) with a fixed frame length of slot allocation. In , Packet Reservation Multiple 2 m sec. used for the uplink (UL) and the downlink Access with Dynamic Allocation (PRMA/DA) (DL) channel. Fig. 1 illustrates the frame structure MAC protocol adopts dynamic allocation algorithm for the uplink channel. The channel bit rate is 4.9 in order to resolve the contention situation quickly Mbps and the data slot size is 53 bytes. The number and avoid the waste of bandwidth that occurs when of slots per frame is 24 slots. The uplink frame is there are several unneeded request slots. However divided into control and data transmission periods, the drawback is that this protocol does not use mini- each consisting of integer number of slots. Slots slots for the access request. In  the use of assigned for control purpose are further subdivided piggybacking information from VBR connection into four control mini-slots with each mini-slot improves the slot allocation for VBR traffic and accommodating reservation mini-packet. enhances the overall protocol performance. The current paper introduces a quantitative comparison of three proposed MAC Protocol for Ubiquitous Computing and Communication Journal 1 Control period Data Transmission period Figure: 1 the frame structure. In the uplink channel, control slots provide a where, communication mechanism for a mobile station to send a reservation request during the contention TF: Frame duration (2 m-sec). phase of the connection. The data slots are provided Tint: Average inter-arrival time of ABR data message with contention-free mechanism during the data (100 m-sec). transmission phase. An uplink control packet is sent whenever a mobile station needs to inform the base Int: largest integer value. station with its traffic characteristics and source Contention period is set to a constant number of status. control mini-slots and this number is chosen to Feedback for the uplink control packets is sent in the satisfy the required QoS for voice traffic. The downlink control packets. contention process is divided to four stages: 2.2 Contention Access Scheme First Stage: When the connection becomes active it The first and second protocols use the same randomly selects one of the 4 contention scheme and the length of the control subsequent frames to send its request period is dynamically adjusted as a function of during the contention period. contention traffic load. The control mini-slots are Second Stage: If the connection exhibits collision in used by the mobile stations to send their reservation the first stage it randomly selects one requests in contention mode using slotted Aloha of the 3 subsequent frames to send its protocol. To reduce the access time of real-time request during the contention period. connection, which greatly affects the QoS of the real-time services, we separate the control mini-slots Third Stage: If the connection exhibits collision in assigned to real-time and non real-time connections. the second stage it randomly selects The number of control mini-slots assigned to real- one of the 2 subsequent frames to time and non real-time connections is adaptively send its request during the contention allocated with the collision status. The total number period. of uplink control mini-slots ranges from 4 to 12 Fourth Stage: If the connection exhibits collision in mini-slots. A priority is given to real-time the third stage it sends its request in connections by assigning their control mini-slots every frame until the base station first according to the number of collisions occurred successfully receives its request. in the previous frame. In the third protocol, the control period is further In every stage the connection randomly select one of divided into contention and polling periods. Control the available contention mini-slots in the selected slots assigned in the control period are further frame to send its reservation request. If the subdivided into four control mini-slots, some of them connection request is correctly received during any used as contention mini-slots and the others used as stage the connection exit from the contention process polling mini-slots. A fixed number of control mini- and the base station periodically allocate slots to the slots are allocated for contention and polling access. connection until the end of talk-spurt. The contention mini-slots are used by voice The described contention process aims to reduce the connections to send their reservation requests in contention load during the contention period in each contention mode at the beginning of talk-spurt, while frame, increase the probability of successfully the poling mini-slots are used by ABR connections accessing the network, decreasing the probability of to send their buffer length status to the base station. collision, reduce the access delay time and at the The number of polling mini-slots are chosen such same time minimize the number of used contention that the polling period will be less than or equal to mini-slots and utilize them efficiently. Decreasing the average inter-arrival time of ABR data message the number of available frames for selection in each (100 m-sec). subsequent stage aiming to reduce the access delay Number of polling mini-slots ≥ int (number of ABR time of the connections and hence reduce the cell users * (TF/Tint)). loss probability. Ubiquitous Computing and Communication Journal 2 2.3 Traffic Integration Strategy 4.1 Dynamic Allocation TDMA MAC Protocol for Wireless ATM Networks As different wireless ATM services share the 4.1.1 Slot Allocation Algorithm for Voice same resources, an effective interaction between the traffic allocation algorithms is needed to maximize the The voice connections have the higher priority. utilization efficiency of the shared resources. In the At the beginning of a talk-spurt, the mobile sends a first and second protocol , the voice connections control packet. When the base station knows that the have the highest priority and the VBR connections connection becomes active the base station have the next higher priority. The ABR connections periodically allocates slots to the connection until have the lowest priority. the end of talk spurt. At the end of the talk-spurt, the In the third protocol, the available transmission mobile sets a flag in the last voice packet to inform slots are assigned first to active voice connections, the base station that the connection is no longer then a minimum assigned slots are allocated to ABR active. traffic, then VBR traffic slots are allocated, and 4.1.2 Slot Allocation Algorithm for VBR traffic finally, the remaining slots are distributed between VBR connections have the next highest priority. ABR connections according to the buffer length of They only contend (send a control packet) at the each connections. session beginning. Next, all the control information is piggybacking on the data packets, which reduces 3 SOURCE MODELS the contention over the real-time mini-slots. At the base station, a token pool of certain size is 3.1 Voice Source Model introduced for each VBR connection. Tokens are A voice source generates a signal that follows a generated at a fixed rate that is equal to the mean pattern of talk-spurts separated by silent gaps. A cell rate. A token is removed from the speech activity detector can be used to detect this corresponding pool for every slot allocated to the pattern. Therefore, an ON/OFF model can describe a connection voice source: the source alternates between the ON After slot allocation for voice connections the base state where the source generates packets at rate 8 station allocates one slot for each VBR connection kbps, and the OFF state where no packets are to send one of their cells and also to piggyback the generated. Durations of talk-spurts and silent gaps current traffic parameter (e.g. buffer length, cell are modelled as exponential distributions with mean delay) of the connection. Then the base station values of 1 and 1.35 sec, respectively. allocates slots for each connection .The number of If a voice packet is not sent within its maximum slots allocated for a connection is the minimum of transfer delay (MTD), it should be dropped The the buffer length and the number of tokens in the MTD is set to be 16 m-sec. pool such as; 3.2 VBR Source Model Nv= min (Av, Bv) . The source rates are modelled as truncated where Gaussian distribution between (128 – 384 kbps) Nv : number of slots allocated for the VBR with mean rate of 256 kbps. The rate of the source connection. varies every 33 m-sec (the duration of image frame) Av : number of tokens in the pool. and the MTD of the VBR packet is set to be 50 m- Bv : number of the packets in the mobile sec. station buffer. Each connection cannot send greater than 12 cells in 3.3 ABR Source Model the frame. Within the frame, priority is given to the It resembles a data source with messages of connection with minimum time-of-expiry to send certain length. The length of the message is their cells earlier. exponentially distributed with mean 2 k bit, and the 4.1.3 Slot Allocation Algorithm for ABR traffic inter-arrival time between messages is negatively The base station records the buffer length exponential distributed with mean of 100 m-sec. status of each connection using the control The MTD of the ABR packet is set to be 6 sec. information transmitted by the mobile. When a message arrives at a mobile, it sends the number of 4 BANDWIDTH ALLOCATION ALGORITHM packets in the new message either piggybacked to a data packet or in a control packet. The bandwidth allocation for uplink Like VBR connections a token pool is introduced transmission is only considered since the downlink for each ABR connection. ABR connections have transmission can be scheduled in the same manner lower priority than voice and VBR connections. The as in a wired ATM switch. number of slots allocated for a connection is the minimum of the buffer length and the number of tokens in the pool such as; Ubiquitous Computing and Communication Journal 3 Na= min (Aa , Ba). introduced for all VBR connections. Tokens are where, generated at a fixed rate that is equal to the mean Na : number of slots allocated for the ABR cell rate per connection multiplied by the number of connection. VBR connections. A token is removed from the Aa : number of tokens in the pool. corresponding pool for every slot allocated to any Ba : number of the packets in the mobile VBR connection. The cell delay is piggybacking on station buffer. the data packets. The connection with higher number of tokens in The number of slots allocated for a VBR connection its pool sends their cells earlier within the frame. If depends on the cell delay and the number of token there are remaining slots inside the frame, the base in the pool such as; station allocates them fairly between ABR and VBR Nvj=int ( Kv* ( Dvj/Dv)) connections. where Nvj : number of slots allocated for the 4.2 An Intelligent MAC Protocol for next connection number j generation Wireless ATM Networks Kv : number of tokens in the pool 4.2.1 Slot Allocation Algorithm for Voice and Dvj : delay time of the last transmitted cell CBR traffic from connection number j The voice connections have the highest Dv : total cell delay of all VBR connections priority. At the beginning of a talk-spurt, the mobile sends a control packet to inform the base station that 4.2.3 Slot Allocation Algorithm for ABR traffic the connection become active. The base station records the buffer length status At the base station, a token pool is introduced for of each connection using the control information each active voice connection and each token is transmitted by the mobile. When a message arrives increased by a fixed amount equal to Tv every frame at a mobile, it sends the number of packets in the to indicate the number of cells generated in the new message either piggybacked to a data packet or mobile station buffer and decreased by one for every in a control packet. slot allocated to the corresponding connection. Then Like VBR connections one token pool is introduced the voice connections are arranged according to the for all ABR connections. ABR connections have content of its token and slots are allocated to the lower priority than voice and VBR connections. The connection with higher value in its token first. At number of slots allocated for an ABR connection the end of the talk-spurt, the mobile sets a flag in the depends on the buffer length and the number of last voice cell to inform the base station that the token in the pool such as; connection is no longer active. Naj= Ka* ( Baj/Ba) Tv=Tf /Tp Where where, Naj : number of slots allocated for the Tf : frame duration (2msec). connection number j. Tp : packetization time of the ATM cell of Ka : number of tokens in the pool. voice connection (48m-sec). Baj : number of cells in the buffer of the The number of slots allocated for voice connection connection (buffer length). in each frame should not exceed Lv . Ba : summation of the buffer lengths of all where ABR connections. Lv = number of voice connection*( Tf / Tp). If there are remaining slots inside the frame, the base station allocates them between ABR and VBR The token poll has two advantages: connections such that Ψ % for VBR connections and First: it indicates the number of packets the rest for ABR connections where; generated at the mobile station buffer. D avg Second: it indicates the amount of delay of the Ψ= ( )*100. generated packet. T oe This helps in deciding which voice connection where should send its packet early and leads to reducing Davg : average delay of VBR connections. the average delay of the voice connections. Toe : time of expiry of VBR cells (50 m-sec). 4.2.2 Slot Allocation Algorithm for VBR traffic 4.3 Contention and Polling based Multiple Access VBR connections have the next higher priority. Control with minimum Piggybacking for They only contend (send a control packet) at the Wireless ATM Network session beginning. Next, all the control information 4.3.1 Slot Allocation Algorithm for Voice Traffic is piggybacking on the data packets, which reduces the contention over the real-time mini-slots. At the At the beginning of talk spurt the voice base station, one token pool of certain size is connection sends a reservation request through the Ubiquitous Computing and Communication Journal 4 contention mini-slot. When the base station of VBR allocated slots becomes lower than Vmean to successfully receives the request, it periodically decrease the counter. allocates slots to the connection up to the end of talk • Each VBR connection could not have spurt. At the end of talk spurt the connection set a lower than one allocated slot per frame. one bit flag in the last transmitted cell to indicate that the connection is no longer active. As we suggested before the delay threshold can be set at a fixed value and its value have a significant 4.3.2 Slot Allocation Algorithm for VBR Traffic effect on the allocation process and the achieved Initially the base station allocates one slot for QoS of VBR traffic. During the simulation at fixed each active VBR connection and then broadcast a delay threshold we take its fixed value equals to 0.5 delay threshold value to all VBR connections every maximum CTD of VBR cell (25msec) as an frame. One bit flag is used to indicate the delay appropriate value and evaluate the performance of status of the buffer and is piggybacked to the data the allocation process in this case. packet (cell). Each VBR connection checks its buffer Table 1: Dynamic adjustment of the delay threshold and sets the flag to one when the packet delay exceeds the delay threshold, and to zero when the packet delay is lower than the delay threshold. The Counter Delay Threshold values slot allocation procedures are performed as follow: (m-sec) • The base station increases the assigned slots by one Counter ≤4 15 for a VBR connection each time its 4 < counter ≤ 8 20 packet delay is greater than the delay threshold (piggybacking flag equal to one). 8 < counter ≤12 25 • At the base station a counter is introduced. The 12 < counter ≤15 30 counter incremented by one when the number of 15 < counter ≤18 35 slots allocated for VBR traffic in the frame is greater than Vmean and decremented by one when it is lower 18 < counter ≤ 20 40 than Vmean. 20 < counter 45 where; Vmean: the mean number of cells generated from all VBR connections per frame according to the mean cell generation 4.3.3 Slot Allocation Algorithm for ABR Traffic rate per connection. Polling control mini-slots are used by ABR • The delay threshold can be set to a fixed value or connections to send their buffer length to the base dynamically adjusted to control the slot allocation station. The number of polling mini-slots is selected process for VBR traffic. The counter can be used to such that the polling period should be lower than or dynamically adjust the delay threshold by increasing equal to the inter-arrival time between ABR data the delay threshold value when the counter value is messages (100 m-sec) to enable the base station to increased and decreasing the delay threshold value efficiently monitor the buffer length status of each when the counter is decreased. Since the increase in connection. the counter value indicates the increase in the Initially a minimum number of slots are allocated allocated bandwidth (slots), so we need to reduce it by to ABR traffic. So that, each ABR connection has an increasing the delay threshold value which in turn allocated bandwidth equivalent to 50 % of its decreases the piggybacking and hence decreases the average cell generation rate. The base station number of allocated slots and vice versa. Table.1 controls this minimum assigned bandwidth by shows the dynamic delay threshold values using the maintaining a leaky bucket for every ABR dynamic adjustment. connection. Tokens added to the bucket at constant • When some of the connection reserved slots are not rate equals to 50% of the average cell generation rate. used by the connection for transmitting its packets Every time a slot is allocated to the connection, a (number of generated packets become lower than the token is removed from the bucket. So, in each frame number of allocated slots) the base station release this the connection with non empty leaky bucket has slot and decrement the number of the reserved slots allocated slots equal to the number of tokens in its for that connection in the subsequent frames by one. bucket. After allocating the VBR traffic slots, the remaining slots are allocated to ABR connections. • When the counter becomes greater than the upper ABR connections are arranged according to their limit value (25) the base station release some of the buffer length where the connection with higher reserved slots for the connections that have no buffer length has its required slots allocated first. piggybacking in the previous frame until the number Ubiquitous Computing and Communication Journal 5 5 PERFORMANCE EVALUATION AND of its VBR Resource allocation algorithm is lower than SIMULATION RESULTS the other two protocols. A comparison has been made to evaluate the For ABR traffic, Fig. 4 and Fig. 7 show that the performance of the three proposed protocols using the reduction of data transmission bandwidth of the third same simulation parameters. protocol by 3.85% due to the contention and polling Fig. 2 through Fig. 7 illustrates the performance periods significantly reduces the available bandwidth with integrated traffic. There are 30 voice for ABR traffic which make the cell losses start early connection and 12 VBR connections in the network, before 40 ABR connection and the average cell delay while the ABR connections are added gradually to significantly increases with ABR connections since a the network. All results are presented as a function considerable part of ABR resource allocations takes of the number of ABR connection. place after VBR slot allocation. The first and second For voice traffic, Fig. 2 and Fig. 5 show that a protocols achieve good QoS for ABR traffic but the good QoS is achieved by the three protocols in term of first protocol achieve slightly better performance in cell loss probability (lower than 10-4) and average cell term of cell loss probability and average cell delay. delay (lower than 5 m-sec). it is clear that, At 45 ABR connection the first and second approximately, the first and the second proposed protocol achieve approximately 94% data transmission protocols achieve better performance than the third one throughput and 98.5% total channel utilization while as they use the same contention access scheme, due to providing the acceptable QoS required for each traffic using lower number of control mini-slots for category. For the third protocol, at 36 ABR connection contention access. It is worth to mention that the third 91% data transmission throughput and 98.5% total protocol uses 8.3% of the bandwidth (8 control mini- channel utilization are achieved while preserving the slots) for contention and polling access, while the first required QoS for each ATM traffic type. and second protocol uses 4.45% of the bandwidth (approximately 4 control mini-slots) which make the available data transmission bandwidth for the third protocol lower by 3.85 % than that of the first and second protocol. For VBR traffic, Fig. 3 shows that with low VBR traffic up to 45 connections, the second protocol achieves the best performance in term of cell loss probability as its resource allocation algorithm depends on the cell delay at each connection buffer, so that the connection with higher delay allocated more slots than that with lower delay which leads to reduce the probability that the cell delay exceeds the maximum CTD (cell transfer delay) and then lost. This decreases the cell loss probability. On the other hand this increases the average cell delay that becomes higher than the average cell delay caused by the first protocol as indicated in Fig. 6.The third protocol achieves higher loss probability than the second protocol (Fig. Figure 2: Cell loss probability of Voice connections 3), and the highest average cell delay (Fig. 6) since the as a function of the number of ABR connections (12 dynamic delay threshold adjustment process produces VBR and 30 Voice connection) more average delay than the other two protocols. When the number of ABR connections becomes greater than 45, the offered traffic becomes higher than the available bandwidth. The third protocol achieves the lowest cell loss probability and lower average cell delay than the second one since a considerable part of ABR slot allocation takes place after VBR slot allocation, and the VBR slot allocation is controlled by the value of delay threshold which has upper limit value. So increasing the number of ABR connection has low significant effect on VBR slot allocation. The first protocol achieves the lowest performance in terms of cell loss probability (Fig. 3), while achieves the lowest average cell delay with all number of ABR connections, (Fig. 6). This indicates that the efficiency Figure 3: Cell loss probability of VBR connections as a function of the number of ABR connections (12 VBR and 30 Voice connection) Ubiquitous Computing and Communication Journal 6 The performance with real time traffic is illustrated in With the third protocol the average cell delay lower Fig. 8 through Fig. 11. 12 VBR connections are than with integrated traffic because in the absence of presented while the voice connections are added ABR traffic if there are remaining slots they will be gradually to the system. For voice traffic, Fig. 8 and given to VBR connections. Fig. 10 show that the performance is the same as with A cell loss probability of 10-3 for VBR traffic integrated traffic, where the first and second protocols achieved by the first protocol at 106 voice connection perform better than the third one. For VBR traffic, Fig. (95% channel utilization), by the second protocol at 9 and Fig. 11 show that the second protocol achieves 112 voice connection (97% channel utilization), and by the lowest cell loss probability. The average cell delay the third protocol at 103 voice connection (94.7% of the first and second protocols is low with slightly channel utilization). different values until 112 voice connection (97% channel utilization), after that, the average cell delay of the second protocol become significantly higher since the efficiency of its resource allocation algorithm in reducing cell loss probability results in increasing the average cell delay. The third protocol has lower cell delay than that with integrated traffic because in the absence of ABR traffic, any remaining slots will be given to VBR connections. Figure 6: Average cell delay of VBR connections as a function of the number of ABR connections (12 VBR and 30 Voice connections). Figure 4: Cell loss probability of ABR connections as a function of the number of ABR connections (12 VBR and 30 Voice connection) Figure 7: Average cell delay of ABR connections as a function of the number of ABR connections (12 VBR and 30 Voice connections). Figure 5: Average cell delay of Voice connections as a function of the number of ABR connections, (12 VBR and 30 Voice connections). Ubiquitous Computing and Communication Journal 7 These results indicate that the second protocol has the protocol uses the lowest piggybacking overhead. For most efficient VBR slot allocation algorithm then the ABR traffic, the reduction of data transmission first protocol and finally the third protocol, but the bandwidth of the third protocol by 3.85% reduce the third protocol uses the lowest piggybacking overhead. available bandwidth for ABR traffic which make the cell losses and the average cell delay significantly increases with higher values than the other two protocol. Finally the three proposed protocols achieve very high channel utilization of approximately 98% for the wireless ATM channel while respects the required QoS of multimedia ATM traffic types. Figure 8: Cell loss probability of Voice connections as a function of the number of Voice connections (12 VBR Connection). Figure 10: Average cell delay of Voice connections as a function of the number of voice connections (12 VBR connections). Figure 9: Cell loss probability of VBR connections as a function of the number of Voice connections (12 VBR Connection). 6 CONCLUSION We have presented an extensive performance comparison of three proposed MAC protocols to highlight the merits and demerits of each of them. For voice traffic, a good QoS achieved by the three Figure 11: Average cell delay of VBR connections as a protocols. But, the first and second proposed function of the number of voice connections (12 VBR protocols achieve better QoS than the third protocol connection) while using lower number of control mini-slots for contention access. For VBR traffic, the results indicate that the second protocol has the most efficient VBR slot allocation algorithm then the first protocol and finally the third protocol, but the third Ubiquitous Computing and Communication Journal 8 REFERENCES On Circuits & Systems (The 46th IEEE International MWSCAS), December 2003,  D. Raychandhuri and N. D. Wilson, “ ATM Cairo, Egypt. –based transport architecture for multi-  Sami A. EL-Dolil and Mohammed Abd Elnaby services wireless personal communication " Contention and Polling Based Multiple networks“ IEEE J. Selected. Areas in Access Control with Minimum Piggybacking Communication, vol.12, no.80, pp. 1401- for Wireless ATM Networks, " 21st National 1414, Oct. 1994. Radio Science Conference (21st NRSC’ 2004), March 16-18, Cairo, Egypt.  J.F. Frigon, H.C.B.Chan and V.C.M. leung “Dynamic reservation TDMA protocol for wireless ATM networks “ IEEE J. Selected Areas in Communication, vol.19, no.2, pp 370-383, Feb.2001.  N.Passas, S. Paskalis, D.Vali and L.Merakos “Quality-of -Service – Oriented Medium Access Control for wireless ATM networks “ IEEE Communication Magazine, pp. 42-50, 1997.  J.Sanchez, R. Martnez, and M.W.Marcellin “ A survey of MAC Protocols Proposed for Wireless ATM “ IEEE Networks, pp. 52-62, Nov. 1997.  H. Liu, U. Cliese and L. Dittmann “ Knowledge-based Multiple Access Protocol in Broadband Wireless ATM Networks “ 50 th Veh. Tech. Conf., pp. 1685-1689, Sept. 1999 . Amesterdam, the Ntherlands.  S. Lee, Y. Song , D. Cho , Y. Dhong and J.Yang “ Wireless ATM MAC layer Protocol for Near Optimal Quality of Service Support” GLOBE COM. 98 Sydney, Australia, pp. 2264-2269, Nov. 1998 .  Y.Kwork and K.N.Lau “A quantitative Comparison of Multiple Access Control Protocols for Wireless ATM “ IEEE Transaction on Vehicular Technology, vol.50, no.3, MAY 2001.  R. Steele and L. Hanzo, Mobile Radio Communications, Wiley and Sons, New York, 1999.  J. C.Chen , K.M.Sivalingam and R.Acharya “ Comparative Analysis of Wireless ATM Channel Access Protocols ” Baltzer Journals, Sept. 1997.  Sami A. EL-Dolil and Mohammed Abd Elnaby " Dynamic Allocation TDMA MAC Protocol for Wireless ATM Networks" Proc. of the Twentieth National Radio Science Conference (20thNRSC’ 2003), March 18-20, Cairo, Egypt.  Sami A. EL-Dolil and Mohammed Abd Elnaby " An Intelligent Resource Management Strategy for Next Generation WATM Personal Communication Network," The Proc. of the 46th IEEE International Midwest Symposium Ubiquitous Computing and Communication Journal 9
Pages to are hidden for
"35ubicc 35"Please download to view full document