Keywords

1 Introduction

To ensure further growth of the IPTV (Internet Protocol TeleVision) market, it is one of the big problems to solve that the IPTV service heavily depends on the service providers. They gather subscribers, aggregate contents, provide the IPTV platform, and make the charge for the services. As the interests in supporting IP communication tremendously increase, a variety of IP-based real time multimedia application services have been developed rapidly in residential and enterprise communication markets because of their attractive service enhancements and cost savings. For the future All-IP based convergence networks and systems in providing high quality, useful quality-guaranteed techniques with considering multimedia and multi-user are researched to support pervasive communication services. As the general quality factor in ubiquitous environment, environmental noise impediment is inevitable in case of that hardware- or software-based communication system located in anywhere and we make phone-call in anytime. Research on noise reduction/speech enhancement can be traced back to about 40 years ago [1–4]. The noise reduction techniques are used to a wide range of applications such as communication, automatic speech recognition, and sound source localization systems today [5–8].

We think that the Open IPTV ought to be beneficial to service, contents, and terminal providers, and subscribers most of all. The important requirements of our Open IPTV design are as follows. Subscribers are able to enjoy every service providers’ contents freely and cheaply with competitions. Service providers maintain their priority to profit from their customers. Thus, we make the specific service overlay environment for the Open IPTV called the Neighbored Garden.

And then, we are able to perceive that in the Open IPTV environment the providers’ main interest is the management of EPG (Electronic Program Guide). Because EPG influences the subscribers to watch the intended VODs, and the actions is directly connected with the profit of the firm.

In the Open IPTV, the service selection problem is very important, because there are numerous overlapped contents and providers. In the service selection problem, the policy [9] is very useful to make the priority of contents [10]. The existing paper about the policy in the walled garden IPTV mainly proposed QoS (Quality of Service) and charge management [10]. And the service provider needs to record the location of subscriber in the service space to manage the authentication, and the billing with the service profile [11] in the Open IPTV.

The monitoring component is essential to our framework to search for the list of the same services and the server condition variables. And the monitoring component can detect contexts [12, 13], so this component is also able to the context awareness and the service managements services for adaptation [14, 15].

The proposed structure is based on the ITU-T IPTV NGN Framework with non-IMS, and similar with the service overlay network [16, 17]. And the NGSON (Next Generation Service Overlay Network) shows the overlay structure.

The EPGs (Electronic Program Guide) must be selectable by subscribers for the IPTV openness [18]. The various and specialized EPG services of several providers such as the EPG for children, adult, etc. will promote user experiences. It’s similar with the internet portal service or the IPTV personalized EPG [19, 20].

2 Media Signal Sharing Scheme with Noise Reduction

The Noise reduction is one of the significant research factors because of degrading the speech through IP network. A clean speech signal \( d(n) \) which is a zero-mean clean speech signal, an additive noise v(n) which is a zero-mean noise process, and recent output of an observation signal as the noisy speech signal at the discrete time sample \( y(n) \) is given by

$$ y(n) = d(n) + v(n) $$
(1)

We propose an optimized Wiener filter method using estimated SNR (Signal-to-Noise Rate) ratio for speech enhancement. The signal power spectrum is computed for the windowed signal using the 256-FFT. Based on the VAD decision, the noise power spectrum is updated only for non-speech intervals in the Wiener filter design. For speech intervals, the last noise power spectrum is reused. And the speech power spectrum is estimated by the difference between the noise power spectrum and speech power spectrum. The designed Wiener filter coefficients in the frequency domain are transformed into the time-domain ones by the IDCT (Inverse Discrete Cosine Transform). The noise is suppressed by the convolution sum between the impulse response of the proposed an optimized Wiener filter which is estimated by SNR measuring and the noisy speech. In the proposed optimized Wiener filter, the frequency response is given by

$$ W(k) = \frac{{\zeta^{\alpha } (k)}}{{1 + \zeta^{\alpha } (k)}}, \, 0 < \alpha \le 1 $$
(2)

and \( \zeta (k) \) is defined as

$$ \zeta (k) = \frac{{P_{d} (k)}}{{P_{v} (k)}} $$
(3)

where k is the frequency bin, \( \zeta (k) \), \( P_{d} (k) \), and \( P_{v} (k) \) are the SNR, the speech power spectrum, and the noise power spectrum, respectively.

The filtering can be controlled by the parameter \( \alpha \). As \( \alpha \) increases, \( \zeta^{\alpha } (k) \) also increases for \( \zeta (k) \) greater than one, while \( \zeta^{\alpha } (k) \) decreases for \( \zeta (k) \) less than one. Therefore, the signal is more strongly filtered out to reduce the noise for small \( \zeta^{\alpha } (k) \). On the other hand, the signal is more weakly filtered with little attenuation for large \( \zeta^{\alpha } (k) \). To analyze the effect of \( \alpha \), we evaluate the performances for \( \alpha \) values from 0.1 to 1.0. We can adaptively select the optimal \( \alpha \) according to the estimated SNR by a logistic function. The logistic function is trained to decide the optimal \( \alpha \) for the estimated SNR. The logistic function used in this paper can be expressed as

$$ p(SNR) = Min + \frac{2(Max - Min)}{{1 + e^{{(\left| {n - 1} \right|/\beta )}} }} $$
(4)

Because the shape of the logistic function changes with the variation of \( \beta \), if the maximum and the minimum values of the logistic function are fixed, we should find the appropriate \( \beta \).

The appropriate \( \beta \) value is decided by the simple gradient search algorithm. At the first iteration, for the initial \( \beta \) value, the corresponding \( \alpha \) as the output of the logistic function is calculated with the estimated SNR as the input of logistic function at each frequency bin for each frame. The average spectral distortion \( J \) for all frames is measured with the log spectral Euclidean distance defined as

$$ J = \frac{1}{N}\sum\limits_{i = 0}^{N - 1} {\frac{1}{L}\sqrt {\sum\limits_{k = 0}^{L - 1} {\left[ {\log \left| {X_{{ref_{i} }} (k)} \right| - \log \left\{ {\left| {X_{{in_{i} }} (k)} \right|W_{i} (k)} \right\}} \right]^{2} } } } $$
(5)

where i is the frame index, N is the total number of the frames, \( k \) is the frequency bin index, \( L \) is the total number of the frequency bins, \( \left| {X_{{ref_{i} }} (k)} \right| \) is the spectrum of the clean reference signal, and \( \left| {X_{{in_{i} }} (k)} \right|W_{i} (k) \) is the noise-reduced signal spectrum after filtering with the designed Wiener filter. At the second iteration, \( \beta \) is updated by the simple gradient search procedure. The average spectral distortion is measured with the new logistic function defined by the updated \( \beta \). Until the termination condition is satisfied, the iteration is repeated. Finally, \( \beta \) is decided after the final iteration.

3 Scenario for Service Overlay Multimedia Framework

In the design of the Service Overlay Multimedia framework in the service overlay control we mainly point two things. One is that subscribers can select any EPG in the network, and the other is that the final decision of service selection depends on the subscriber’s policy.

The framework design with the ITU-T IPTV NGN framework with non-IMS, and it is composed of the Application Function, the Service Control Function, the End-User Function, the Content Delivery Function, and the Data Transport Function.

We mainly re-designed the Service Control Function with 5 functional blocks: the Service Overlay Control Block (SOC) mainly connects other service providers to provide the Open IPTV services, and the User Profile Management Block (UPM) manages the user profiles, they are not shared with other service providers.

The Service Monitoring Block (SM) works for the gathering information in the service overlay network, the Service Policy Decision Block (SPD) is the brain to make decision with personal policy, and the Service Routing/Negotiation Block (SRN) control the transmission on the overlay network. Figure 1 and Table 1 shows the configuration and the capability of the framework blocks.

Fig. 1
figure 1

Framework for open IPTV

Full size image
Table 1 Capability of the Blocks
Full size table

The SOC, the UPM, and the SRN work on the service overlay network and the SOC is the Gateway to connect with the other service providers so the UPM and the SRN work through the SOC but we omit the SOC in the sequence diagram of the SM, or the SRN. Figure 2 shows the overlay network for the Open IPTV network.

Fig. 2
figure 2

Service overlay multimedia network

Full size image

The framework has advantages to works on the service overlay network and be adaptive to the personal policy compared with ITU-U NGN framework with non-IMS.

4 Search EPGs and Transmit EPGs

The Open IPTV subscriber can select the EPG which is provided in the network, and receive the EPGs list. When the subscriber request EPG list, the SM collect the EPG list on the service overlay network. And after a EPG is selected by subscriber, the SRN request the transmission of EPG. Figure 3 shows the sequence diagram of the EPG selection scenario.

Fig. 3
figure 3

Sequence of EPG selection

Full size image

5 Evaluation

We implemented our test system of the Open IPTV based on the service scenarios. Our test system used the ASP.NET framework, and Java for the XML (eXtensible Mark-up Language) Web Service (Tables 2, 3).

Table 2 Main functions for EPG selection
Full size table
Table 3 Main functions for VoD selection
Full size table

6 Conclusion

In this paper, the performance evaluation of speech quality confirms that our proposed shared scheme outperforms more efficiently than the original algorithm in the speech codecs. The performance results in this paper have established and confirmed that our proposed shared packet loss-aware robust noise reduction scheme using estimated input-SNR outperforms more efficiently than the existing method, which control noisy signal considering speech distortion problem with packet loss. To sum up, our proposed method allows and shows the strength with the improved performance results for the noisy signal reduction at the same post-processing time while the noisy signal is removed over packet-based IP networks. Also, we proposed the service overlay media framework. And we show that subscribers are able to select the provider freely on the service overlay network. So we believe that the neighbored garden Open IPTV would be a new chance to not only the customers but also the providers.