Section on scheduling very hard to follow, mainly because of a num- ber of grammatical errors and lack of organization. Examples pre- sented are confusing and not too clear. Very difficult to follow. One example is section 2.4. Is this section referring to Figure 4? If so, please state. Where did C1 = 4, D1 = 10, etc. came from? In contrast, section about synchronization is a lot better. Though some points are still hard to follow, they can be understood a lot better than the scheduling section with careful reading. Examples presented in this section are pretty clear, (very nice example about semaphores in section 3.3). One other comment: real-time systems vs. real-time operating systems? The paper seems to imply that they are synonymous, but are they? -------- Grammar was a problem through much of the scheduling part of the paper. This detracted significantly from the presentation. Sometimes new terminology was not defined or clear from its usage - i.e. "perfectly preemptible", "P1 and P2" (period or priority?) on top of page 8, "socketed" on page 1. If it is relevant terminology than it should be explained. If it's irrelevant then effort should be made to replace it with something equally meaningful but understandable within the context. This way, you can avoid digressing from the point without having the reader trying to find where in the paper the term is defined (when it isn't defined in the paper anywhere). A large portion of the scheduling section seemed to come from Liu and Layland (from 1973) even though other scheduling papers are cited. Timeliness is a concern. The impression a reader could draw is that this field of research is not active. Why isn't preemptive scheduling also an issue for static scheduling? It is only mentioned in the dynamic scheduling section but is implied in the static scheduling section with the words: "getting a CPU slice". Introducing preemption before static and dynamic scheduling would make this a little clearer. Had trouble with the graphs. The X-axis was time, and the Y-axis was priority, but in which order? Label both axes to make this clearer. The paragraph on Java at bottom page 9 is a little confusing. It seems like you are proposing that embedded OSs should or are commonly written with Java which is (i believe) not the case. Upon rereading, the paragraph is there to illustrate a point about synchronization involving an event queue. Maybe you could introduce the problem before this analogy for a more consistent logical flow. The central problem of each section was clearly presented which gave logical continuity through each section. Also, both sections concurred with the overall theme of embedded real-time systems. -------- This paper was well-organized, but there are numerous grammatical mistakes that make it hard to read. With regards to the topic however, their topic is very key as real-time OSs are indeed an important field today. They provided good background/intro to the area, and discussed a wide variety of synchronization and scheduling methods. The diagrams are hard to understand. Overall, I thought they brought up some very good ideas - but it was just hard to understand what they were trying to say sometimes. -------- I liked the real-time example of controlling a nuclear reactor. Although giving definitions as an introduction is not the best aproach, I think a broader introductionary aproach should be used. I liked the easy to read difinitions but the landry aproach on p.2 and p.3 was not the best. I feel incorporating the definitions using transitions is best. Also great examples of scheduling algorithims. Good Interupt transparent sync. model. -------- - Clearity: Some parts is not clear, especially section 2.3. It needs less details(equations) but more explanations. There is a small mistake with the equation in 2.3.1, I guess. - The style of referring to source papers is not well-formed. - The focus should be more on describing the techniques specific to embedded real-time systems, rather than general cases. - An embedded real-time system can be analysed as an example for both sheduling and synchronization purposes. - A brief explanation before the subsections would be better. (e.g after 2.2, 2.3, 2.3.2 etc.) -------- I like the nuclear reactor example. The presentation and writing styles in the two halves of the paper, scheduling and synchronization, are drastically different. The scheduling section seemed to deal more with fine details while synchronization dealt with the overall big picture. The paper could be improved if it flowed better. Also, the paper probably should talk about the research systems more, and discuss the cross-cutting issues between them. -------- Many of the explanations in this paper were clear, particularly the synchronization methods. The synchronization discussion could have benefited from some analysis to bring the different methods together. I had a little more trouble with the scheduling discussion. This was a little disturbing to me, as I have read a bit about real-time scheduling algorithms (my OS paper was also on real-time OS's). In terms of writing style, this section had a number of distracting grammar mistakes, among other problems. I think this section would have been clearer if the different techniques had been compared. I think that this paper is excellent in its introduction of the problem and the topic of scheduling tasks. Its biggest failure is in only the descriptions of the scheduling paradigms and algorithms. -------- Very complete coverage of the issues and concepts pertaining to the topic. All terms were well defined. Good analysis and comparison of scheduling mechanisms. It would have been helpful to present a similar analysis/comparison for synchronization. Well written with good style. -------- Scheduling and Synchronization in Embedded Real-Time OS By Khushu, Simmons What authors observe is that scheduling and synchronization problems in real time operating system aren't easy, due to both the time constraints and the lack of resources than conventional system. And, by providing a background idea about scheduling and synchronization which are core in traditional OSes, they raise what the problem is when it comes to real-time environment and describe the proposed solutions. Unfortunately, I don't think they have tried to argue or prove something from surveying in this area. Even though they issues somewhat hot problems in especially hard-real time system, I'm afraid to say that it is not interesting to read this paper because of lack of challenging. -------- Some figures unclear, need labeling, more explanation. Introduction mentions hard real-time, soft real-time, distinction not mentioned later in describing systems. -------- Scheduling and synchronization are both important topics in OS but I can't see any reason why both of them should be included in a single survey. Either of them can fill a whole paper. For scheduling part, it is very general and the speciality of embedded system is not fully represented. The algorithms are very old. As far as I know, scheduling problem is well studied in theoretical computer science. I think the authors can just focus on synchronization part. I choose the scores for the following reasons Import: 4 Either scheduling or synchronization is not a good topic since I am not persuade there are much difference in real-time embedded system. In generally systems, these two topics are well-documented. Novelty: 4 This paper only describes the existing algorithms and approaches. And those algorithms are quite old. Quality: 5 The scheduling algorithms are well presented. Overall: 5 They studied a lot of paper and finish it by only two persons. -------- This paper was carefully considered, and a relatively applicable and active area of OS research, but was not well presented. I found many grammatical errors detracting from the soundness of the paper, indicating failure to properly proofread the work. In addition, while the concepts of synchronization and scheduling are crucial in real-time operting systems, I did not believe they were tied together concretely. The paper should have ellaborated further on how the two relate and influence the design decisions of one another. In addition, it seemed like the 'embedded' nature of the OS was of secondary importance, while this would seem a serious restriction on the design and implmentation of an operating system. Further elaboration, plus directions for new research, would make this paper more concrete and interesting to read. -------- The paper gives a good overview of the real-time systems but I found the provided examples and figures hard to decipher. I wish authors have spent more time explaining them. Also certain parts of the paper sounded clumsy and were hard to follow. -------- There is enough in the paper to do cover scheduling. Perhaps the impact of the paper could have been magnified by just one topic. It is almost as if the reader is distracted by a second subject rather than concentrating on the first. On page 5, in 2.3.1 the sentence "...remaining tasks monotonically in the order." is incomplete. In the order of what?