A Bandwidth Reservation Multiple Access Protocol for Wireless ATM Local Networks

A bandwidth reservation multiple access scheme(BRMA) is proposed to resolve contention and assignbandwidth among multiple users trying to gain access toa common channel such as in mobile users contending for resources in an ATM-based cellular networkor a wireless local area network (LAN) with shortpropagation delays. The protocol is best suited tosupport variable-bit-rate (VBR) traffic that exhibits high temporal fluctuations. Each mobile user isconnected end-to-end to another user over virtualchannels via the base station that is connected to thewired ATM B-ISDN network. The channel capacity is modeled as a time frame with a fixed duration.Each frame starts with minislots, to resolve contentionand reserve bandwidth, followed by data-transmissionslots. Every contending user places a request for data slots in one of the minislots. If therequest is granted by the base station through adownlink broadcast channel, the user then startstransmission in the assigned slot(s). The number ofassigned slots varies according to the required qualityof service (QoS), such as delay and packet lossprobability. A speech activity detector is utilized inorder to indicate the talkspurts to avoid wastingbandwidth. Due to its asynchronous nature, BRMA is ratherinsensitive to the burstiness of the traffic. Since theassignment of the minislots is deterministic, therequest channels are contention-free and the data channels are collision-free. Hence, in spite ofthe overhead (minislots) in each frame, BRMA provideshigher throughput than Packet Reservation MultipleAccess (PRMA) for the same QoS, especially for high-speed systems. A better delay performance is alsoachieved for data traffic compared to Slotted Alohareservation-type protocol PRMA. In addition, BRMAperforms better in terms of bandwidth efficiency thanthe conventional TDMA or the Dynamic TDMA, wherespeech activity detectors are very difficult toimplement.     

Coordinated Resource Assignment in a Fixed Wireless System

In this paper, we study the interferencemanagement and resource allocation problem in apacket-switched, fixed, broadband, Time DivisionMultiple Access (TDMA) wireless system. These systemshave the potential to provide high-speed data access to the homeand therefore are of current interest. The systemcapacity is limited by the co-channel interference,which arises from both intercell and intracelltransmissions. The intracell interference management andresource allocation, which significantly impacts thesystem performance, is the focus of this study. The goalis to design efficient resource allocation schemes to combat intracell interference so as toimprove the system capacity and service quality. Weaddress the following issues: (1) estimating the optimalcapacity of the system; (2) proposing resourceallocation schemes with reasonable complexity that areclose to optimal; and (3) proposing a new class of powercontrol schemes that are suited to intracellinterference management. Two resource allocation schemes are proposed: Max-Min Scheduling Protocol(MMSP) and Revisited Max-Min Scheduling Protocol(RMMSP). We show via simulations that these schemescombined with the new power control provide close tooptimal performance.     

Performance of a Dynamic TDMA System for Mobile Wireless Access

A dynamic TDMA system can utilize voice activityand allow the integration of voice and data traffic.This can be achieved by allocating frequency channelsand time slots on demand. In this approach, upon the arrival of a talkspurt or a data packet,the base station is requested to assign a time slot foreach transmission. Message requests and assignments ofmobile users are carried over a Control channel, while the voice and traffic are transmittedover a Traffic channel. Time slot assignments are madefrom a pool of Traffic channels. A numberof slots in the pool will be shared by voice and data, with voice having priority over data, andthe remaining will be used by data only. Voice slots arereserved for the duration of the talkspurt whereas datapackets are assigned on a per-slot basis. Data packets can be buffered whereas voicetraffic can only tolerate limited delay beyond whichtalkspurts will be clipped off. The Control channeluplink access is based on Slotted Aloha so that mobile users have autonomous access to base stations.This paper presents the performance of the dynamic TDMAsystem outlined here. The analysis aims at assessing thecapacity gained by using voice activity and voice/data integration, in terms of theimpairments introduced to voice quality (e.g., speechclipping and/or delay) and the delays to data packets.The analysis has been based on a discrete time Markov model operating on a frame-by-frame basis thatprovides the joint distribution of the number of activevoice and data users in the system. The analysis alsoevaluates the delays of message requests via the uplink control channel. In evaluating theclipping probability, we combine the impact of both theaccess delays at the control channel as well as theunavailability of time slots in the pool. Performance results indicate that the capacity gain mayexceed 80% and the speech clipping can be kept below 1%.Also, data packets may be transmitted with limiteddelays even when all capacity is allocated for voice users. The proposed approach may be used toenhance the capacity of the existing TDMA cellularsystems and to provide integration of voice and dataservices.     

A Constant Propagation Algorithm for Explicitly Parallel Programs

In this paper, we present a constant propagation algorithm for explicitly parallel programs, which we call the Concurrent Sparse Conditional Constant propagation algorithm. This algorithm is an extension of the Sparse Conditional Constant propagation algorithm. Without considering the interaction between threads, classical optimizations lead to an incorrect program transformation for parallel programs. To make analyzing parallel programs possible, a new intermediate representation is needed. We introduce the Concurrent Static Single Assignment (CSSA) form to represent explicitly parallel programs with interleaving semantics and synchronization. The only parallel construct considered in this paper is cobegin/coend. A new confluence function, the -assignment, which summarizes the information of interleaving statements between threads, is introduced. The Concurrent Control Flow Graph, which contains information about conflicting statements, control flow, and synchronization, is used as an underlying representation for the CSSA from.