FIFO segmentation as a substitute to linked lists

Petreas, FC; Stasinopoulos, GI

dc.contributor.author	Petreas, FC	en
dc.contributor.author	Stasinopoulos, GI	en
dc.date.accessioned	2014-03-01T01:28:24Z
dc.date.available	2014-03-01T01:28:24Z
dc.date.issued	2008	en
dc.identifier.issn	1434-8411	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/18839
dc.subject	List processing	en
dc.subject	Memory management	en
dc.subject	Shared memory	en
dc.subject.classification	Engineering, Electrical & Electronic	en
dc.subject.classification	Telecommunications	en
dc.subject.other	Computer simulation	en
dc.subject.other	Cost accounting	en
dc.subject.other	Data transfer	en
dc.subject.other	Input output programs	en
dc.subject.other	List processing	en
dc.subject.other	Memory management	en
dc.subject.other	Shared memory	en
dc.subject.other	Storage allocation (computer)	en
dc.title	FIFO segmentation as a substitute to linked lists	en
heal.type	journalArticle	en
heal.identifier.primary	10.1016/j.aeue.2007.05.006	en
heal.identifier.secondary	http://dx.doi.org/10.1016/j.aeue.2007.05.006	en
heal.language	English	en
heal.publicationDate	2008	en
heal.abstract	We present a technique for dynamic allocation of memory to different queues. Our approach lies between two well-known techniques, linked lists and the relocation of data for preserving reasonable contiguous areas of available memory space. However, we avoid the main drawbacks of both. We never allocate more than two segments of contiguous memory per flow and thus avoid the proliferation of pointers and associated memory fragmentation. Moreover, we never relocate data already admitted into memory. We thus offer a considerably simpler implementation, amenable to hardware realization. The price to pay is also twofold. We cannot guarantee total exhaustion of the available memory before overflow occurs and we can only implement the first in first out (FIFO) discipline for the flows hosted in memory. In fact, we exploit the defining feature of a FIFO: data having arrived first will also leave first, spending in memory as little time as possible and giving the opportunity for rearranging memory allocation to the advantage of future arrivals. Two segments per flow are sufficient to achieve, over time, complete memory reallocation without data movement. Simulation results exhibit this 'refreshing' feature of FIFO, as brought to light by our proposed scheme. (C) 2007 Elsevier GmbH. All rights reserved.	en
heal.publisher	ELSEVIER GMBH, URBAN & FISCHER VERLAG	en
heal.journalName	AEU - International Journal of Electronics and Communications	en
dc.identifier.doi	10.1016/j.aeue.2007.05.006	en
dc.identifier.isi	ISI:000255886100004	en
dc.identifier.volume	62	en
dc.identifier.issue	5	en
dc.identifier.spage	356	en
dc.identifier.epage	364	en