The first impression of an e-reader is very important to the user. An easy-to-use interface and a friendly home page are essential. For now, being lightweight and having a big storage capacity are the main advantages of e-readers. Moreover, most e-readers are web based; it is very convenient to download e-books anytime and anywhere. A user can also easily access to the feedbacks and reviews about a particular e-book from other users.
Usually, the user likes to have other readers’ opinions to guide him/her to choose. On the contrary, many students have raised the problem that most of the online libraries only have few e-books and most of them are too specific.
The suggested improvements in term of features and functionalities are annotations and an easier way to navigate in the e-books. Many students also point out the fact that they choose e-reader to benefit from the fact that e-books can have links, embedded dictionary,
search tools that don’t exist in paper books. From the performance point of view, connectivity is very important. That is, to be able to find the right e-book to read anywhere, anytime, fast Internet connection is a must. Moreover, the general responsiveness of the e-reader is a key issue. Many users complain about the waiting time of chapter and image loading while reading. Sometime we may need to wait several minutes to read a DRM protected e-book due to data decryption. To summarize, the user experience of e-reader still can be improved. For the features and functionalities, e-reader needs to make the most of its electronic support. The performance issue is even more important, especially for DRM protected e-books. The last chapters of this these will deal with the particular issue of chapter loading.
Chapter 4
Chapter Preloading Mechanism for E-Reader
In e-reader applications, the content files (chapters) of an e-book are downloaded/decrypted to the buffer of an reader for reading. For DRM protected e-books, to prevent the content from illegally copied, only the current chapter for reading is decrypted and loaded into the buffer of the e-reader. When the user jumps to another chapter, he/she has to wait until the new chapter is loaded. To reduce the waiting time, we propose in this chapter a preloading mechanism that preloads a small number of chapters to significantly improve the user experience and still assure the required protection level for the e-book.
4.1. Chapter Loading Performance Issues
In Chapters 2 and 3, our studies on e-readers focused on their features and functionalities.
Performance of e-reader is seldom investigated in the literature, which can be measured by two indices: the content loading time and the device resource consumption [20].
The loading time influences the user experience directly. With long loading time, the user cannot read smoothly through the e-book. The resource consumption includes the CPU usage of the device and the memory buffer that holds the e-book content in the device. We
have studied the CPU usage issue in chapter 2, and will not elaborate on it here. In this chapter, we consider the loading time and the buffer size issues. E-book reading can be implemented in two alternatives:
Offline reading is the most common implementation in readers [13, 10], where an e-book is downloaded offline into the e-reader as a zipped file. To read the e-e-book, it has to be unzipped and the content files are loaded into the buffer individually. Each content file typically represents one chapter of the e-book.
Online reading is typically implemented for comics and journals reading [21, 18]. The e-book chapters are stored in the server of a dedicated website or online library, instead of downloading them into the file directory of the e-reader. The chapters to be read are delivered directly from the server to the buffer of the e-reader through, e.g., mobile Internet.
In both alternatives, chapter loading is the most time and resource consuming procedure.
The easiest way to resolve this issue is to preload the whole book into the buffer of the e-reader [6]. However, this method is not applicable for DRM protected e-books, where the content files are encrypted to prevent the whole e-book from being illegally copied [8, 15].
That is, to deliver DRM protected e-book content, either from the server or the zipped-file directory of the device, only the current chapter for reading is loaded into the buffer.
Therefore, a malicious user can potentially copy at most one chapter of the e-book.
However, it means that each time the user jumps to another chapter, he/she has to wait until the new chapter is loaded.
Based on the reading behavior and the e-book characteristics, this chapter presents a method to improve chapter loading by optimizing the e-reader buffer allocation. We apply on DRM protected e-books a chapter preloading mechanism that loads a limited number of unzipped chapters into the buffer of an e-reader. We will carefully select the number of loaded chapters to limit the damage caused by a malicious user, while significantly improve the reading experience.
The next section presents the chapter preloading mechanism that loads a limited number of unzipped/decrypted chapters into the buffer. The last section evaluates the performances of this mechanism and concludes chapter.
4.2. Chapter Preloading Mechanism
This section proposes a preloading mechanism for e-book. We use a multi element ring buffer for chapter loading [23], where each buffer element holds one chapter of the e-book.
Our method preloads into the buffer a small number of chapters having the highest probabilities to be read.
Let N be the number of chapters of a given e-book. We first rank all chapters according to its Most Likely Reading Order (MLRO) as shown in Figure 4.1
Figure 4.1 Most Likely Reading Order (MLRO)
Chapter number i 1 … n - j … n - 1 n n +1 … n + j … N MLRO rank kn(i) … 2j+1 … 3 1 2 … 2j …
MLRO is established based on common reading habits, i.e. the reading order is sequential and chapters close to the current reading one have higher probabilities to be read. Let n be the current chapter (1 ≤ n ≤ N), and kn(i) be the rank of any chapter i with respect to n. In Figure 4.1, the gray-colored box indicates chapter n, which has the rank kn(n)=1. Rank 2 is assigned to the next chapter in the sequential order: kn(n+1)=2, rank 3 is assigned to the previous one, i.e. kn(n-1)=3, and so on. We note that user’s “backward” reading may not exactly follow the above ranking. However this ranking does capture the trend in our experiments at university library. For the description purpose, we define two operators ⊕ and as follows:
n⊕ j = (n + j mod N) and n j = (n - j mod N)
Note that if (n + j)>N, then n ⊕ j = n + j - N; and if (n - j)<1, then n j = n – j + N.
To accommodate the ⊕ and the operators on the range [1,N]={1,2,…,N}, we define a notation to represent a positive integer range:
In (4.1) and (4.2), we call a the left-hand side of a,b , and b the right-hand side of a,b .
While reading an e-book, the user can jump from chapter n to another chapter n*. If the jump is to the right of chapter n within half of the e-book (i.e. ⎢⎣N / 2⎥⎦ chapters), then it is a forward jump. Similarly, if the jump is to the left of chapter n within half of the e-book (i.e. N / 2
(
⎡⎢ ⎤⎥ −1)
chapters), then it is a backward jump. That is, a forward jump impliesFrom the above description, the MLRO ranking can be mathematically defined as follows.
Definition 1. Let n* be the next chapter and j=|n*-n| be the number of chapters the user jumps forward or backward. For an e-book of size N, the MLRO ranking of chapter n*
When j=0, n*=n and kn(n*)=1, which is considered as a backward jump.
Directly from Definition 1, we have the following fact.
Fact 1: If j=|n*-n| is the number of chapters the user jumps, then
Note that MLRO can be adjusted based on other reading habits. In this chapter, we evaluate the reader performance based on MLRO in Definition 1. Performance of e-reader with other chapter rankings can be similarly investigated and will not be presented.
If the current chapter is changed, MLRO re-ranking is triggered, where the re-established ranking is centered on the new chapter using Definition 1. Note that the user may jump to a page in the same chapter. In this case, MRLO remains the same. In the preloading mechanism, we always load the chapters with the smallest MLRO ranks into the buffer as described in the following definition.
Definition 2. Let B be the buffer size and n the current chapter. Then chapter i is stored in the buffer if 1 ≤ kn(i) ≤ B.
Based on Definition 2, chapter i is in the buffer if
Note that if B ≥ N, the whole e-book is held in the buffer and there is no chapter loading issue. If B < N, when the user jumps to a new chapter, some or all of the loaded chapters in the buffer must be replaced. The number of chapters to be replaced is derived in the following Theorem.
Theorem 1. Suppose that B<N. For a jump from chapter n to chapter n*, let j =|n*-n|. Let NL(j) be the number of chapters to be replaced in the buffer. Then
NL( j)=
Proof: Note that for every jump, j ≤ N/2, and in the hypothesis, we do not need to consider the case when j > N/2. Suppose that B is odd (For B is even, the proof is similar and will not be presented). From (4.4), chapter is in the buffer where
S= n B− 1
For the demonstration purpose, we assume that S satisfies (1), i.e.
1≤ n − B− 1
Since we use a symmetrical ring buffer, assumption (4.5) does not compromise the generality of the proof. Assume that the user jumps j chapters forward. After the jump, chapter is in the buffer where
Scenario 1: j <B. We first consider the right-hand side of S. Since j <B, the left-hand side of S* in (4.6) is equal or smaller than the right-hand side of S in (4.5), and chapters i in the r portion of S satisfy
(n+ j) − B− 1
Now, we consider the left-hand side of S. If l ≠ 0, it implies that and
NL( j)= N − B (4.9)
Since l ≠ 0, (4.6) must satisfies (4.2); in other words,
(n+ j) B− 1
Furthermore, the right-hand side of S* in (4.6) is equal or larger than the left-hand side of S in (4.5). That is,
Suppose l ≠ 0, then (4.10) must be satisfied. That is j >N-B. Since j ≤ N/2 (definition of a forward jump), we have
N− B < N
2 ⇒ B > N
2 (4.13)
Since j ≥ B, (4.13) implies that j >N/2 which contradicts with the fact that j ≤ N/2.
Therefore, Scenario 2 holds if and only if B ≤ j ≤ N–B and l=0,
NL( j)= B (4.14)
From (4.12) and (4.14), the hypothesis is proved. Note that for a backward jump, the proofs for both Scenarios 1 and 2 are similar except that n +j is replaced by n – j.
Q.E.D.
4.3. Performance Evaluation
Two output measures are considered for the performance evaluation of the chapter preloading:
The reload probability p is the probability that the next chapter n* is not in the buffer.
That is, with probability p, the user needs to wait for the loading of chapter n* and cannot read smoothly through the e-book.
The loading cost CL is the expected number of chapters that are replaced in the buffer at each MRLO re-ranking due to a chapter jump.
Note that in a practical e-book application, the buffer size cannot be too large, and for the practical exercise, it suffices to consider B<N/2. In this case, the loading cost also reflects the “security cost”. If more chapters are loaded into the buffer, the penalty of being illegally copied is higher.
This section shows how to select appropriate B values to yield good p and CL performance.
We derive p and CL assuming that the probability distribution of e-book chapter reading follows Zipf’s law. This law was initially used to predict the frequency of appearance of words in a corpus in English language. It is an empirical law stating that the frequency of
any word is inversely proportional to its rank in the order of frequency [17]. More generally, Zipf's law predicts, out of a population of N elements, the frequency of the element of rank k to appear. The parameter s describes the behavior of the population (to be elaborate). The Zipf’s probability of the element of rank k is
(4.15)
In this study, the e-book chapters are the population under consideration, where N is the number of chapters of a given e-book. The parameter s describes the reading behavior, and more specifically, the locality of the read chapters. A larger s means that the user is more inclined to stay in the same chapter; on the contrary, a small s means that there is a higher probability to jump to chapters further away from the current one. Zipf’s law gives the probability of any chapter to be read depending on its MLRO rank given in Definition 1.
For a jump from chapter n to chapter n*, n* needs to be loaded in the buffer for immediate reading if kn(n*)>B. Therefore, from (4.15), the reload probability p is expressed as
(4.16)
A low reload probability p means that most of the time, the user can read through the e-book without waiting.
Now we derive CL. Since B<N/2, Theorem 1 can be simplified as
Let C1 be the loading cost corresponding to forward jumps. In this case, j≤ N / 2⎢⎣ ⎥⎦. From probability p. On the contrary, from (4.19), a larger buffer size B implies a higher loading cost CL. The goal of this study is to select an appropriate buffer size B to balance against p and CL. In doing so, we define a net cost C that considers the effect of both p and CL as follows:
C= p +αCL
where α is a factor that weights the importance of CL against p.
Figure 4.2 shows the reload probability p as a function of B/N and s, where N=20. The figure indicates the intuition that p decreases as B/N increases. We observe a non-trivial result that there is a knee point (the bullets in Figure 4.2) such that before this point, p significantly decreases as B/N increases. This phenomenon implies that after this knee point, selecting a larger buffer size does not improve p performance. Such knee effect is more significant for a larger s. Figure 4.2 also indicates that to achieve the same p performance, B/N deceases as s increases.
Figure 4.2 Reload Probability for N=20
Figure 4.3 illustrates that to achieve the same p performance, a larger B/N is required for a smaller N. That is, a larger portion of e-book needs to be stored in the buffer for an e-book of small volume. Similar to Figure 4.2, a knee point can be determined on every p curve.
Figure 4.3 Reload Probability for s=2
Figure 4.4 shows the loading cost for N=20 under various s values. We observe that CL
increases as the locality parameter s or the e-book size B/N increases; this effect becomes less significant with large s values.
Figure 4.4 Loading cost for N=20
The net cost for s=2 is illustrated in Figure 4.5. For α=0.6, we observe a minimum value on each C curve for N=50 and N=100 respectively. In both cases, this minimum occurs when B=3 (B/N=0.06 for N=50 and B/N=0.03 for N=100). By increasing the factor α, the protection of the e-book is higher and the minimum occurs for a smaller B value (for α=0.8, B=2). We note that the B/N value of the minimum is always smaller than that of the knee point for p.
Figure 4.5 Net Cost for s=2
The above performance study indicates that by slightly increasing the buffer size (i.e.
slightly compromising the protection of the e-book), the user experience is significantly improved. In the example in Figure 4.5, by buffering two more chapters (i.e., B/N increases from 0.02 to 0.06 for N=50 and α=0.6), we can reduce the reload probability p from 0.38 to 0.17 (Figure 4.3), which is a 56% improvement. Our chapter preloading mechanism provides a method for e-reader designers to select the appropriate buffer size according to the requirements in term of user experience and e-book protection.
Chapter 5
Conclusion and Future Works
This chapter concludes our work about the performance evaluation and user experience of e-readers and presents briefly the future directions of this study.
5.1. Contribution
E-reader has introduced a brand new experience of reading compared to the traditional paper book. The development of e-readers is still in its growing phase. The users need to know and to familiarize with this new technology; at the same time, e-reader designers try their best to fulfill the users’ needs. Generally speaking, e-reader features and functionalities are designed to imitate the reading experience of paper books. Moreover, additional features are added to make the most of the electronic support, such as connectivity, links and storage capability. Our studies showed that for university students, these features are the ones that make them prefer e-books to paper books.
The performance of e-reader is a key issue under consideration to improve user experience.
It is obvious that with long response time and high resource consumption, no one will use e-readers. From the performance evaluation in this thesis, we observe that for now, e-book and chapter loading performances are rather poor. Moreover, for DRM protected e-books
these loading performances are even worth caused by data decryption. Then, we described a chapter preloading mechanism to improve performance. In the presented mechanism, we showed that by preloading a small number of chapters, we could significantly improve the user experience and still assure the required level of e-book protection.
5.2. Future Works
This thesis has pointed out the weaknesses and the possible improvements for e-readers regarding the user experience. In terms of features, we found out that the e-reader should benefit more from its electronic support. We will focus our future work on the design of additional features such as internal links, embedded dictionary or search tools to bring to the user an innovative and easier way of reading and learning.
Moreover, in the chapter preloading mechanism, strong assumptions on the reading behavior were made to define the MLRO. However, we note that user’s reading may not exactly follow this ranking. This ranking depends on the user’s own reading behavior as well as the type of the e-book content. Our future work will also consider the design of reading models suitable for specific type of e-books. The final goal is to implement this mechanism on actual e-readers.
Biography
[1] Apple - Instruments User Guide - Performance Analysis Tools, Apple Inc. 2009.
Available: http://developer.apple.com/iPhone/library/documentation/Developer Tools/ Conceptual/InstrumentsUserGuide/Introduction/Introduction.html
[2] Apple - iPhone - Technical Specifications, 2010. Available:
http://www.apple.com/iphone/specs- 3g.html.
[3] Chong, P.F., Lim, Y.P. and Ling, S. W. “E-Book Design Preferences: A Case Study”, International Symposium on Information Technology, 2008.
[4] Chuang, C.Y., Wang, Y.C. and Lin, Y.B. “Digital Right Management and Software Protection on Android Phones”, IEEE 71st Vehicular Technology Conference (VTC2010-Spring), 2010.
[5] Cripps, T. “If iPhones and Androids: Redefining the Smartphone and Other Devices,” Ovum Report, March 2009.
[6] Freda: the Free Ebook Reader. Available: http://www.turnip.demon.co.uk/jim/freda/
[7] Google - Google Books. Available http://books.google.com/
[8] He, G. “Analysis of E-Book Security”, Article in T-110.501 Seminar on Network Security, 2001. Available: http://www.tml.tkk.fi/Studies/Tik110.501/2001/papers/
he.pdf
[9] Hillesund, T. and Jon, E. Noring “Digital Libraries and the Need for a Universal Digital Publication Format,” the Journal of Electronic Publishing, vol. 9, no. 2, summer 2006.
[10] Kindle-Amazon: Wireless Reading Device. Available:
http://www.amazon.com/Kindle-Wireless-Reading-Device-Display/dp/B00154JDAI
[11] International Digital Publishing Forum (formerly Open eBook Forum). Available:
http://www.idpf.org/
[12] International Digital Publishing Forum Specifications. Available:
http://www.idpf.org/specs.htm
[13] Lexcycle Stanza: a Revolution in Reading. Available: http://www.lexcycle.com/
[14] Lexcycle - Lexcycle Has Been Acquired by Amazon.com! Lexcycle, April 27 2009.
Available: http://www.lexcycle.com/lexcycle_acquired_by_amazon
[15] Liu, Q., Safavi-Naini, R. and Sheppaed N.P. “Digital Rights Management for Content Distribution”, Proceedings for the Australasian Information Security Workshop Conference on ACSW Frontiers 2003.
[16] Moore, L.M. “At Your Leisure: Assessing Ebook Reader Functionality and Interactivity”. Available: http://www.uclic.ucl.ac.uk/distinction-projects/2009-Moore.pdf
[17] Newman, M. E. J. “Power Laws, Pareto Distributions and Zipf’s Law”
Contemporary Physics, vol 46, pp. 323-351, 2005.
[18] Questia: Trusted Online Research. Available: http://www.questia.com/
[19] Rave, P. “How to Approach the “Base of the Pyramid”: A Business-Strategy and
[19] Rave, P. “How to Approach the “Base of the Pyramid”: A Business-Strategy and