A Scheme for Lossy Joint Source-Channel Coding

Web Published:

            Researchers at Princeton University have developed a solution for the simultaneous lossy transmission of a natural source (such as images, audio and video) over a one-to-many (broadcast) channel under a quadratic distortion fidelity criterion.    Princeton is currently seeking an industrial collaborator to commercialize this technology.


Conventional approaches to multimedia streaming over heterogeneous digital networks consist of establishing individual streaming sessions from the server to each user.  This may be very inefficient when many users wish to receive the same content.  A definite advantage of analog broadcasting systems is the capability to send simultaneously the same signal to a potentially unlimited number of receivers, with possibly different reconstruction quality that depends on channel conditions.


This new technology employs the multicasting of independently parallel Gaussian sources over a binary erasure broadcast channel.  Multiresolution embedded quantization and layered joint source channel coding schemes are used in order to serve simultaneously several users at different channel capacities.  The convex nature of the rate-distortion function, computed by means of a reverse water-filling, allows the solution of relevant convex optimization problems corresponding to different performance criteria.  Layered joint source channel codes are constructed based on the concatenation of embedded scalar quantizers with binary rateless encoders.


         It is anticipated that this technology will be useful in advanced techniques for wireless multimedia delivery, such as video on demand applications.


Patent protection is pending.


            For more information please contact:


                        Michael R. Tyerech
                        Office of Technology Licensing 
                        Princeton University
                        87 Prospect Avenue
                        Princeton, NJ 08544-0036

                        (609) 258-6762

                        (609) 258-1584 fax


Patent Information:
For Information, Contact:
Chris Wright
Licensing Associate
Princeton University
H. Vincent Poor
Maria Fresia
Giuseppe Caire
Ozgun Bursalioglu Yilmaz