Time-Efficient Read/Write Register in Crash-prone Asynchronous Message-Passing Systems

Atomic registers are certainly the most basic objects of computing science. Their implementation on top of an n-process asynchronous message-passing system has received a lot of attention. It has been shown that t

This paper investigates time-efficient implementations of atomic read-write registers in message-passing systems where the number of readers can be unbounded. In particular we study the case of a single writer, multiple readers, and S servers, such that the writer, any subset of the readers, and up to t servers may crash. A recent result of Dutta et al. [3] shows how to obtain fast implementations in which both reads and writes complete in one communication round-trip, under the constraint that the number of readers is less than St - 2, where t 2 . In that same paper the authors pose a question of whether it is possible to relax the bound on readers, and at what cost, if semifast implementations are considered, i.e., implementations that have fast reads or fast writes.This paper provides an answer to this question. It is shown that one can obtain implementations where all writes are fast, i.e., involving a single round-trip communication, and where reads complete in one to two communication rounds under the assumption that no more than t 2 servers crash. Simulated scenarios included in this paper indicate that only a small fraction of reads require a second communication round. Interestingly the correctness of the implementation does not depend on the number of concurrent readers in the system. The solution is obtained with the help of non-unique virtual ids assigned to each reader, where the readers sharing a virtual id form a virtual node. For the proposed definition of semifast implementations it is shown that implementations satisfying certain assumptions are semifast if and only if the number of virtual ids in the system is less than St - 2. This result is proved to be tight in terms of the required communication. It is shown that only a single complete two-round read operation may be necessary for each write operation. It is furthermore shown that no semifast implementation exists for the multi-reader, multi-writer model.

Hybrid ARQ is being considered as a potential method for improving the performance of WCDMA in UMTS. We compare different hybrid ARQ techniques, namely the incremental redundancy type II hybrid ARQ schemes and type I hybrid ARQ schemes with and without soft combining. The emphasis is at high data rates and downlink shared channel (DSCH). Both link and system simulations are used for comparison. The hybrid ARQ schemes are shown to achieve higher throughput and capacity than the conventional ARQ schemes currently specified in the 3GPP WCDMA standard. Also, the combining of packets can cut the tail of the delay distribution which reduces the required maximum number of transmissions. However, the complexity increase due to physical layer buffering in the mobile station is too high if the round trip delay is large.

An experimental performance analysis of MAC multicast in 802.11b networks for VoIP traffic

It is shown that implementing a practical self-stabilizing sliding window protocol requires a bound on the maximum delay or maximum memory of the communication channel involved. This motivates using communication channel models that incorporate a delay or memory bound. For such models, two new ARQ protocols are presented that self-stabilize by using 1 bit of overhead in each transmitted message. The protocols operate like selective repeat ARQ, except that when a fault places them in an incorrect (unsafe) state, the additional bit in the protocol messages allows automatic recovery. Following a transient fault, the bounded delay protocol stabilizes within four round-trip times. The bounded memory protocol stabilizes after sending at most 2(K+n) messages, where K the is maximum number of messages that can be stored in one direction on the channel, and n is the window size of the sender.

A fault-tolerant and stabilizing simulation of an atomic register is presented. The simulation works in asynchronous message-passing systems, and allows a minority of processes to crash. The simulation stabilizes in a pragmatic manner, by reaching a long execution in which it runs correctly. A key element in the simulation is a new combinatorial construction of a bounded labeling scheme accommodating arbitrary labels, including those not generated by the scheme itself.

We consider the problem of reducing the control cost associated with handling connections (or flows) in multimedia networks such as ATM or an Internet supporting RSVP. This cost lies in processing and storage; it is directly related to connection setup latency and to the maximum number of connections that a switching node can handle. We advocate a systematic grouping of connections inside the transit network in order to reduce connection awareness at intermediate switching points. In the ATM case, this means using dynamic virtual path connections as trunks between ATM access nodes. We analyze the requirements and issues posed by such an architecture. As a first step towards a complete solution, we have developed a signaling prototype that shows how virtual path trunks can be setup or released without additional round trip delays.

High speed downlink packet access (HSDPA) is specified in the 3GPP to cope with the increasing demand of high-speed Internet-based multimedia services. HSDPA is based on techniques such as base station scheduling, adaptive modulation and coding, hybrid automatic repeat request, and advanced receiver design. This work presents system-level simulation results on the HSDPA performance to reveal the effects of user equipment (UE) capabilities such as the maximum number of multicodes and receive diversity. The results show that a 15- and 10-multi-code UE and the receive diversity significantly improve the cell capacity. Experiments using the HSDPA test-beds are conducted to investigate the true HSDPA performance. Furthermore, TCP/IP-level simulation results show the application-level throughput, i.e. end-user throughput, in the HSDPA system. The results indicate that reduction of the round trip time, a higher uplink data rate, and low uplink block error rate are required to achieve the high TCP-level throughput.

This paper presents a new algorithm that reduces multivalued consensus to binary consensus in an asynchronous message-passing system made up of n processes where up to t may commit Byzantine failures. This algorithm has the following noteworthy properties: it assumes t<n/3 and is consequently optimal from a resilience point of view, uses On2 messages, has a constant time complexity, and does not use signatures. The design of this reduction algorithm relies on two new all-to-all communication abstractions. The first one allows the non-faulty processes to reduce the number of proposed values to c, where c is a small constant. The second communication abstraction allows each non-faulty process to compute a set of proposed values such that, if the set of a non-faulty process contains a single value, then this value belongs to the set of any non-faulty process. Both communication abstractions have an On2 message complexity and a constant time complexity. The reduction of multivalued Byzantine consensus to binary Byzantine consensus is then a simple sequential use of these communication abstractions. To the best of our knowledge, this is the first asynchronous message-passing algorithm that reduces multivalued consensus to binary consensus with On2 messages and constant time complexity measured with the longest causal chain of messages in the presence of up to t<n/3 Byzantine processes, and without using cryptography techniques. Moreover, this reduction algorithm uses a single instance of the underlying binary consensus, and tolerates message re-ordering by Byzantine processes.

RNGAVXLIB: Program library for random number generation, AVX realization

While lots of consensus algorithms have been proposed for crash-prone asynchronous message-passing systems enriched with a failure detector of the class Ω (the class of eventual leader failure detectors), very few algorithms have been proposed for systems enriched with a failure detector of the class P (the class of perfect failure detectors). Moreover, (to the best of our knowledge) the early decision and stopping notion has not been investigated in such systems. This paper presents an early-deciding/stopping P-based consensus algorithm. A process that does not crash decides (and stops) in at most min(f+2, t+1) rounds, where t is the maximum number of processes that may crash, and f the actual number of crashes (0≤ f≤ t). Differently from what occurs in a synchronous system, a perfect failure detector notifies failures asynchronously. This makes the design of an early deciding (and stopping) algorithm not trivial. Interestingly enough, the proposed algorithm meets the lower bound on the number of rounds for early decision in synchronous systems. In that sense, it is optimal. The paper then presents an original algorithm that implements a perfect failure detector in the Theta model, an interesting model that achieves some form of synchrony without relying on physical clocks. Hence, the stacking of these algorithms provides an algorithm that solves consensus in the Theta model in min(f+2, t+1) communication rounds, i.e., in two rounds when there are no failures, which is clearly optimal.

We consider caching in cellular networks in which each base station is equipped with a cache that can store a limited number of files. The popularity of the files is known and the goal is to place files in the caches such that the probability that a user at an arbitrary location in the plane will find the file that she requires in one of the covering caches is maximized. We develop distributed asynchronous algorithms for deciding which contents to store in which cache. Such cooperative algorithms require communication only between caches with overlapping coverage areas and can operate in asynchronous manner. The development of the algorithms is principally based on an observation that the problem can be viewed as a potential game. Our basic algorithm is derived from the best response dynamics. We demonstrate that the complexity of each best response step is independent of the number of files, linear in the cache capacity and linear in the maximum number of base stations that cover a certain area. Then, we show that the overall algorithm complexity for a discrete cache placement is polynomial in both network size and catalog size. In practical examples, the algorithm converges in just a few iterations. Also, in most cases of interest, the basic algorithm finds the best Nash equilibrium corresponding to the global optimum. We provide two extensions of our basic algorithm based on stochastic and deterministic simulated annealing which find the global optimum. Finally, we demonstrate the hit probability evolution on real and synthetic networks numerically and show that our distributed caching algorithm performs significantly better than storing the most popular content, probabilistic content placement policy and Multi-LRU caching policies.

Efficient Asynchronous Simulation of a Class of Synchronous Parallel Algorithms