Other Free Encyclopedias » Online Encyclopedia » Encyclopedia - Featured Articles » Contributed Topics from P-T » Peer-to-Peer Streaming - Streaming P2P Architectures, Streaming Process, Peer-to-Peer System Operation

Challenges for Peer-to-Peer Streaming

system systems network service

P2P systems are designed to distribute the workload and network traffic among the peers and take advantage of the computing and storage resources of each individual peer. There are two aspects to this approach. One the positive side, a P2P system is very scalable and can potentially serve a very large streaming community where the network and processing load will be a significant challenge for a centralized system. The drawback of P2P systems is that because of the dynamic and unpredictable nature of peers a more complicated, fully distributed protocol is required to constantly maintain the system and recover from errors. A lack of centralized control also introduces difficulties for the administration and security of P2P systems. Below we list a few of the challenges commonly encountered in a P2P streaming system.

Quality of Service (QoS)

The quality of service (QoS) of a streaming system usually refers to the end-users experience. Criteria may include the smoothness of the display, the frequency of visual distortions, and the startup latency from session initiation to the onset of the display. QoS requirements depend on the type of P2P streaming systems. For example, users can tolerate a relatively longer delay in a non-interactive streaming system such as on-demand movie watching. This is in contrast to the requirements of a live, two-way audio conferencing system in which the delay must be bounded at the millisecond scale.

The fact that a P2P system is connected in a distributed topology introduces some challenges that are usually less relevant for a centralized system. For example, in order to accommodate a large number of members, P2P systems usually build an application-level overlay network among all members. The resulting stream forwarding or processing at the application level increases the end-to-end delay through the additional intermediate hops from source to destination and as a result it is difficult to build a low latency streaming platform using a P2P platform. Some existing work has investigated low-latency P2P streaming. One idea is to distinguish active users who require low latency from passive users who can tolerate longer latency. By clustering the active users logically close together the delay among them can be reduced The remaining challenge is to distinguish active users effectively and automatically.

Dynamics

One of the biggest challenges for all P2P streaming systems is how to provide a reliable service over an unreliable, constantly changing and most likely, heterogeneous streaming architecture. The members of a P2P system are often of different computing power, network bandwidth and network connectivity. Some are connected from behind a firewall and some are connected through a network address translation (NAT) device. Peers may join and leave at any moment, leaving some fraction of the P2P streaming network isolated and disconnected. These dynamics make the construction of a reliable and deterministic streaming service very challenging.

A common solution is to maintain redundant information to recover the lost service. For example, in the AudioPeer system, each peer caches information about a fraction of the other online peers and when there is a disruption, this cached peer list is used to repair the network. Another possibility is to ask the rendezvous point server for help. This approach is easier to implement and the service may be more reliable since the server is usually monitored and maintained professionally. However such a centralized recovery design hinders the scalability of a distributed system and may increase its costs. A hybrid approach that combines the above two designs is often a good compromise.

Security

Starting from the early days when P2P systems were mostly used for file sharing until today’s blooming online audio conferencing systems that employ P2P as the streaming architecture, security has always been a big concern that affects the acceptance of P2P applications. Members in a P2P system are usually untrusted entities and service is received through such peers. This cooperative model opens the door to unfair service distribution (i.e., a peer only consumes services but does not provide any) and abuse. Since most peers computers are not maintained and configured by network security professionals, a P2P network provides an opportunity for hackers and malicious attacks (e.g., injection of bogus content).

Aside from worries that the P2P service could open a back door to intruders, many people are also concerned about the possibility that confidential information can be obtained by the intermediate peers who are used to relay the content form source to destination. Unfortunately many of current paradigms for P2P systems are limited to the authentication phase and scant research has been done in terms of how to assure the integrity and confidentiality of content being delivered. It will be helpful to revise some research proposals in multicast areas and find the appropriate implementations for P2P streaming platform.

Profit model

Despite the popularity and promising advantages of P2P streaming systems, finding a viable business model remains elusive. One of the challenges is to measure the usage of each individual in a distributed system and charge fairly for the services received. Since each peer in the P2P streaming system is acting as both a customer and a provider, it would not be fair to charge the user by the volume of content received without considering her contribution to help relay the content to other peers. It is also quite challenging to monitor the activities of peers even when permission is given. Various P2P streaming systems may require different profit models. One possible revenue stream is to display advertising on a companion website or in the content itself. Another possibility is to charge a fee for add-on services. Creating a fair and efficient subscription model for P2P streaming systems is a practical challenge that needs to be resolved before P2P streaming systems become a mature commercial platform.

 

User Comments

Your email address will be altered so spam harvesting bots can't read it easily.
Hide my email completely instead?

Cancel or