Current admission controls aiming to maximize network utilization create a bias against large-bandwidth calls. This paper proposes a BNP (buffer without preemption) model which will hold a call in a buffer rather than directly reject it when the residual bandwidth is insufficient. No other call can be admitted until this queued call is served. Such a mechanism increases throughput because the buffer can temporarily hold a call, and reduces the bias for largebandwidth calls because all calls will be rejected when the buffer is not empty, even the residual bandwidth is enough for small-bandwidth calls. The novel is compared with other two alternative models, NB (no buffer) and BP (buffer with preemption), and the performance of the models is analyzed by solving a two-dimensional Markov chain. The analytical results indicate that the proposed model can achieve perfect fairness. Meanwhile, some numerical results confirm that the novel approach achieves higher fairness than the two competing models.
Abslmcl-Current admission controls seeking to optimize network uti-Lizatios create a bias against largcbsndnidth & Thb papet pmprather than directly reject it when the resldual bandwidth is insudded. Such a marbanism inem-thmugbput sinee the b a e r cm tompordy bold a d: it ala0 redueer the bias for Iarge-baodnidlb caUa since all E~U I are rejected when the buffet is not empty, wco when the residual bandwidth suffice for smaU-bsndwldtb & The performaneeofthe models Is nnalynd by sdlvlng a multidimendod Markov chain. A mdieed m d d , BNP-W, in r*hieb a raiting lbrahoid h added, is finher oaered to p m c n l an overlong wsit for the queued c a l l Also, diUere.1 thresho l b for ~t i o u s bmdnidtb e d a are set lo maintain the optimal Mm-Anaiytlyucsl nod numelied r a d & demonstrate that Ibis laner model aetu-aUy =hi-optimal fah-, wen when the perf& Mmcu can not be achieved a BW @,taer nitbout preemption) m d d that xili bdd P eau to a bnmer
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.