Innovative Transport Layer Protocols

Stream Control Transmission Protocol (SCTP)

The Stream Control Transmission Protocol (SCTP) is a new transport protocol, existing at an equivalent level as UDP (User Datagram Protocol) and TCP (Transmission Control Protocol), which currently provide transport layer functions to all of the main Internet applications. SCTP has been approved by the IETF as a Proposed Standard RFC 4460. As with TCP, SCTP provides a reliable transport service, ensuring that data is delievered from to the receiver without error and in the same sequence as transmitted. As with TCP, SCTP is a connection-oriented mechanism, meaning that a relationship is created between the endpoints of an SCTP session prior to data being transmitted, and this relationship is maintained until all data transmission has been successfully completed.

Unlike TCP, SCTP provides a number of functions that are considered critical for signaling transport, and which at the same time can provide transport benefits to other applications requiring additional performance and reliability. Two core innovative servces of SCTP are multistreaming and multihoming. For further details about SCTP, see SCTP: An Overview or SCTP for Beginners.

In our lab, we are currently working on the following projects:

Chrome-SPDY over SCTP
As part of their "Let's make the web faster" initiative, Google is experimenting with alternative protocols to reduce the latency when downloading web pages. One of these experiments is SPDY (pronounced "SPeeDY"), an application-layer protocol for transporting content over the web, designed specifically for reducing latency when Chrome, Google’s browser that competes with Internet Explorer and Firefox. For the past decade, Professor Amer and his students in UD’s Protocol Engineering Laboratory have been developing the Stream Control Transmission Protocol (SCTP). SCTP is an innovative transport protocol, existing at an equivalent level as UDP (User Datagram Protocol) and TCP (Transmission Control Protocol), which currently provide transport layer functions to all of today’s Internet applications. As with TCP, SCTP provides a reliable transport service, ensuring that data is delivered from to the receiver without error and in the same sequence as transmitted. As with TCP, SCTP is a connection-oriented mechanism, meaning that a relationship is created between the endpoints of an SCTP session prior to data being transmitted, and this relationship is maintained until all data transmission has been successfully completed. Unlike TCP, SCTP provides a number of functions that are considered critical for signaling transport, and which at the same time can provide transport benefits to other applications requiring additional performance and reliability. Two core innovative services of SCTP are multistreaming and multihoming. Sponsored by a $50K gift from Google, Professor Amer will collaborate with Google researchers by investigating the performance of Chrome-SPDY over SCTP, and comparing results with Google’s experiments using Chrome-SPDY over TCP. The goal is to demonstrate that features of SCTP that are unavailable in TCP can further improve Chrome-SPDY’s latency gains and/or fault tolerance.

HTTP over SCTP Multistreaming
TCP is the traditional reliable transport protocol for HTTP-based network applications. We argue that SCTP's improved features such as multistreaming, and protection against DoS attacks make SCTP an ideal web transport. Recently, we proposed a design for HTTP over SCTP streams, and implemented HTTP/SCTP in the open source Apache web server and Firefox web browser. Our experiments over an emulated network using Dummynet reveal that SCTP's enhanced loss recovery improves web page download times for HTTP 1.1 persistent with pipelined transfers over high latency end-to-end paths. A multistreamed web transport's primary contribution to minimize response times is concurrent rendering, where multiple pipelined objects can be displayed in an interleaved fashion at the web browser. We explore the parameter space where concurrent rendering causes visually perceivable improvements to pipelined objects' response times. Concurrent rendering's dramatic improvements while downloading progressive images can be viewed at

Movie animations comparing HTTP over TCP vs. HTTP over SCTP

Transport Layer Reneging
A TCP or SCTP data receiver informs a data sender about out-of-order arrived data using Selective Acks (SACK [RFC2018]) mechanism. Thus the data sender sometimes can avoid retransmitting SACKed data during loss recovery. SACK information is "advisory"; SACKed data cannot be discarded from the data sender's send buffer until acked by a cumulative ack because it is permitted for a data receiver to discard previously SACKed data, for example, when the operating system is running out of memory. This "sacking and then discarding" is known as reneging. If no reneging occurs, the SACKed data stored in the send buffer is unnecessary. Our research tries to identify if reneging actually happens in practice. We have downloaded publicly available anonymized Internet traces from CAIDA. containing more than 50M TCP flows. For those TCP flows that employ SACKs, we capture the view of SACKed data of the data receiver. We compare each new SACK to the captured view to see if they are consistent. If reneging occurs, a SACK block reported previously will disappear or shrink. Another way to examine reneging is via SNMP. For example, the LINUX operating system has a LINUX-MIB which extends the TCP-MIB, and there are some built-in objects such as LINUX_MIB_PRUNECALLED which will be incremented if reneging occurred. We are working on an extension to SCTP-MIB agent on FreeBSD to report SCTP reneging. Another aspect of this research is to identify TCP stack implementations of major operating systems that have built-in mechanisms for reneging.

Misbehaviors in TCP SACK Generation
While analyzing CAIDA Internet traces of TCP traffic to detect instances of data reneging, we frequently observed seven misbehaviors in the generation of SACKs. For example, in one case, when the cumulative ACK is increased beyond the first SACK block, a data receiver misbehaved and did not selectively acknowledge the second SACK block. These misbehaviors could result in a data sender mistakenly thinking data reneging occurred. With another misbehavior, the worst case could result in a data sender receiving a SACK for data that was transmitted but never received. Our research presents a methodology and its application to test a wide range of operating systems using TBIT to fingerprint which ones misbehave in each of the seven ways. You can download the patch file www.eecis.udel.edu/~amer/PEL/tbit-tests/tbit-update.patch to update tbit-1.0 www.icir.org/tbit/tbit-1.0.tar.gz to include new TBIT tests to detect misbehaving TCP stacks. Also you can find the collected trace files for 27 operating systems tested at www.eecis.udel.edu/~amer/PEL/tbit-tests/tbit-tests-SACK-Generation-Misbehavior.tar. Measuring the performance loss due to SACK generation misbehaviors is outside the scope of our research; the goal is to document the misbehaviors so they may be corrected. One can conclude that the handling of SACKs while simple in concept is complex to implement. [Read More]

WWW over SCTP
We are modifiying the popular open source browser Firefox, to support HTTP over SCTP. The current activity is focused on adding support for SCTP to the Netscape Portable Runtime API (NSPR), which is used to support Firefox. The goal is to provide support on all NSPR supported platforms and have the modifications accepted into the NSPR distribution. Subsequent work will focus on modifying Firefox to run HTTP over SCTP, using the SCTP API calls added to NSPR.

Non-Renegable Selective Acknowledgments (NR-SACKs)
For both TCP and SCTP, out-of-order data is acknowledged with Selective Acks(SACKs). Currently for both protocols, SACK information is advisory; SACKs notify a data sender about the reception of out-of-order data, but since a receiver is permitted to later discard the data (ie, reneg), the sender must keep a copy of the data in its buffer for possible retransmission until the data is cumulatively acked. SCTP and TCP treat out-of-order data differently. In TCP, out-of-order data CANNOT be delivered to a receiving application until all data up to that point is received in-order. In multistreamed SCTP, out-of-order data CAN be delivered to a receiving application if the data is in-sequence within its stream. Similarly SCTP allows an application to specify data as unordered (a feature that is unavailable in TCP), which also can be delivered to the receiving application immediately upon arrival at the transport receiver. Non-Renegable Selective Acknowledgments (NR-SACKs) allow a receiver to inform the sender of delivered data. (Note: Delevered data by definition cannot be reneged!) NR-SACKs thus allow the data sender to remove those data from the sender's retransmission buffer. NR-SACKs can also be used to acknowledge undelivered data that in the case that a receiver guarantees never to reneg. We have specified Internet Draft: 'draft-natarajan-tsvwg-sctp-nrsack' - SCTP Data Acknowledgment with Non-Renegable Selective Acknowledgments (NR-SACKs), and are currently investigating the circumstances under which NR-SACKs improve end-to-end throughput.

Recent projects include the following:

Mobile SCTP
Inter-working of future heterogeneous radio access networks within the concept of all-IP beyond 3G wireless data networks will pose many technical challenges, with mobility management being one of the most important. We survey mobile Stream Control Transmission Protocol (mSCTP) in context of transport-layer handoff management, as one of most relevant schemes being currently studied. We identify the key scenarios, where the protocol can effectively leverage the multihoming feature to enhance handover support. Finding the basic SCTP's failover mechanism inadequate for mobility applications, the main challenging issue is to provide suitable path-transition optimization by considering link-layer support, as well as introducing concurrent multipath transfer (CMT) into mSCTP-based mobility schemes. Our experiments scope not only on providing a set of triggering rules that help to choose when to switch the paths, but also on evaluating the possible gain in making such a decision at the transport layer. Furthermore we compare the gain that can be introduced by using CMT (CMT-PF) in such mobility scenarios See L. Budzisz, R. Ferrus, K.-J. Grinnemo, A. Brunstrom, R. Fracchia, G. Galante, F. Casadevall, Towards Transport-layer Mobility: Evolution of SCTP Multihoming, Computer Communications, 31(5), 3/08

CMT with the Potentially-failed Destination State (CMT-PF)
Recent research on Concurrent Multipath Transfer using SCTP multihoming (CMT) proposed various retransmission policies to minimize the negative impacts of congestion-induced receiver buffer (rbuf) blocking. The proposed retransmission policies do not consider failure-induced rbuf blocking in CMT. We investigated CMT's throughput degradation during path failures, and proposed CMT with a Potentially-failed destination state (CMT-PF) to improve performance. Ns-2 simulation results show that CMT-PF outperforms CMT during permanent and short-term failures. During non-failure scenarios, CMT-PF performs better than CMT when the paths experience asymmetric rbuf blocking conditions. In light of our findings, we recommend that CMT be replaced by CMT-PF in existing and future CMT implementations and RFCs.

Adaptive Failover
SCTP supports multihoming at the transport layer to allow SCTP associations to remain alive even when an endpoint's IP address becomes unreachable. SCTP has a built-in failure detection and recovery system, known as failover, which allows associations to dynamically send traffic to an alternate peer IP address when needed. SCTP's failover mechanism is static and does not adapt to application requirements or network conditions.  We are developing an adaptive failover mechanism. [Read More]

Concurrent Multipath Transfer (CMT)
With the provisioning for multihoming in SCTP, we demonstrate that end-to-end CMT can be performed at the transport layer.  Being better informed than the application layer about the end-to-end paths, the transport layer can perform fine-grain load sharing. We foresee issues in areas such as congestion control and loss detection and recovery. We are investigating (1) end-to-end techniques for shared bottleneck detection to aid in performing correct congestion control during CMT, and (2) algorithms for scheduling traffic on the multiple paths. [Read More]

Multiple File Transfer using SCTP Multistreaming
We identify overheads associated with FTP, attributed to separate TCP connections for data and control, non-persistence of the data connections, and the sequential nature of command exchanges. Solutions to avoid these overheads using TCP place an undue burden on the application.  We modify FTP to use SCTP and its multistreaming service. FTP over SCTP avoids the identified overheads in the current FTP protocol without introducing complexity at the application, while still remaining TCP-friendly. We implemented FTP over SCTP in three ways: (1) simply replacing TCP calls with SCTP calls, thus using one SCTP association for control and one SCTP association for each data transfer, (2) using a single multistreamed SCTP association for control and all data transfers, and (3) enhancing (2) with the addition of command pipelining. Our experiments compared these 3 variations with the classic FTP over TCP. Results indicate significant improvements in throughput for multiple file transfers combining SCTP's feature of multistreaming with command pipelining. More generally, this research encourages the use of SCTPs innovative services, such as multistreaming, to benefit existing and future application performance. [Read More]

SCTP Shim for Legacy TCP Applications
We are modifying the transport layer of TCP- and SCTP-capable freeBSD systems to invisibly translate application-layer calls to the TCP API into corresponding SCTP API calls. SCTP would then replace TCP as the end-to-end transport protocol used for communication between the source and destination. This approach would provide legacy FCS TCP applications the beneficial features of SCTP that are not provided by TCP, such as multihoming's fault tolerance, concurrent multipath transfer, extensive SACK support, and multistreaming.  With an SCTP shim, these features could enhance communications without making any modifications to the legacy TCP applications. 

U.S. Army Collaborative Technology Alliance

At UD, we are investigating SCTP to evaluate its use within army networks. For the period May 2001 until May 2009, the CISC and ECE Departments will be part of a new US Army Research Laboratory (ARL) Collaborative Technology Alliance (CTA) in Communications and Networking. This CTA brings together a total of seventeen university and industrial partners to collaborate in four general research areas: Survivable Wireless Mobile Networks, Signal Processing for Communications-on-the-Move, Secure Jam-Resistant Communications, and Tactical Information Protection. Principal investigators directing UD's research effort are CISC Professors Paul Amer, Errol Lloyd, and Adarsh Sethi, and ECE Professors Gonzalo Arce and Stephan Bohacek.

The Army has allocated $76.3M for the Communications and Networking Alliance; $44.3M for the first five years and $32M for an optional three additional years. Of those totals, the University of Delaware will receive a projected $3.5M for the first five years and a total of $6M if the Army exercises the optional three-year period. These amounts will be equally divided among the CISC and ECE departments. These funds support approximately fifteen graduate research assistants each year.

The Army Research Laboratory's strategy is to exploit commercial technology and expertise by establishing industrial/academic consortia that work closely with ARL scientists and engineers to help fulfill critical Army modernization objectives. UD is one of fourteen Consortium Members that include industrial partners - Telcordia, BBN Technologies, BAE Systems (previously Lockheed-Sanders) , General Dyanamics (previously Motorola), and Network Associates; and university partners - City College of New York, and U. of Maryland, Princeton, Clark Atlanta, Georgia Tech, U. of Minnisota, Johns Hopkins, and Morgan State.  Additional subcontracts go to: Cornell,  U.C. Riverside, Washington U., New Mexcio State, and U. Michigan.

The history of the CTA program actually began in 1996 when ARL initiated a new strategy that focused in-house laboratory research on Army-specific business areas while establishing extramural centers of research in areas where state-of-the-art expertise could be leveraged to satisfy Army technology needs. The combination of government in-house, industry, and academic components striving together for excellence created a new paradigm for Army research - a "federated laboratory". From 1996-2001, UD's CISC and ECE Departments shared $3.1M as part of the ARL's first FEDLAB program called ATIRP - Advanced Telecommunications and Information Distribution Research Program. Considered an overwhelming success, the FEDLAB program was widely recognized and was awarded the Hammer Award for Reinventing Government by then Vice President Al Gore. ARL decided to capitalize on FEDLAB's success by expanding and improving the concept with the creation of Collaborative Technology Alliances.

This CTA grant is significant to the CISC Department, representing the largest single contract the department has ever received. Being part of the winning alliance proposal demonstrates how strongly and how successfully the CISC and ECE departments have collaborated. CISC and ECE have worked closely together ever since they received an NSF equipment grant to establish a joint EE-CIS computing facility in 1985 that connected three DEC VAX machines with one of the world's first Ethernets.

Partial Order Transport Service for Multimedia and Other Applications

A Partial Order Connection (POC) transport layer service/protocol (published as RFC 1693) has been designed. Unlike classic transport services that deliver objects either in the exact order transmitted or according to no particular order, POC provides a partial order service; that is, a service that requires some, but not all objects to be received in the order transmitted. 

Two versions of POC have been proposed: reliable, which requires that all transmitted objects are eventually delivered, and unreliable, which permits the service to lose a subset of the objects. In the unreliable version, objects are more finely categorized into one of several reliability classes depending on their temporal value. 

POC has been formally specified in Estelle and its performance has been investigated via analytic modeling, an OPNET simulation, and an implementation based on a modified TCP. 

Related Publications

 

Authors	Title	Published	Paper
F. Yang	Nonrenegable selective acknowledgments and scheduling for TCP and multipath TCP	PhD Dissertation, CIS Dept, Univ of Delaware, 2015	PDF
N. Ekiz P. Amer	Transport layer reneging	Computer Communications, 52, 10/14, p82-88	PDF
F. Yang P. Amer	Using one-way communication delay for in-order arrival MPTCP scheduling	9th EAI ChinaCom 2014, Maoming, CN, 8/14	PDF
F. Yang Q. Wang P. Amer	Out-of-order transmission for in-order arrival scheduling for MPTCP	PAMS 2014, Victoria, CA 5/14	PDF
F. Yang P. Amer N. Ekiz	A scheduler for MPTCP	22nd IEEE ICCCN 2013, Nassau, BS, 8/13	PDF
N. Ekiz P. Amer F. Yang	Causing remote hosts to reneg	7th IEEE PMECT 2013, Nassau, BS, 8/13	PDF
F. Yang P. Amer	Non-renegable selective acks (NR-SACKs) for MPTCP	PAMS 2013, Barcelona 3/13	PDF
N. Ekiz	Transport layer reneging	PhD Dissertation, CIS Dept, Univ of Delaware, 2012	PDF
F. Yang P. Amer J. Leighton	A methodology to derive SPDY's initial dictionary for zlib compression	CISC Dept Technical Report, 2011	PDF
I. Aydin J. Iyengar P. Conrad C. Shen P. Amer	Evaluating TCP-friendliness in light of concurrent multipath transfer	Computer Networks, 56(7), 5/12	PDF
J. Li M. Manley M. Veeraraghavan R. Williams M. Reisslein P. Amer J. Leighton	A less-is-more architecture (LIMA) for a future Internet	15th IEEE Global Internet Symposium, Orlando, 3/12	PDF
L. Bertaux P. Berthou T. Gayraud P. Amer	Geo-localization to enhance SCTP handover in public transports between satellites and WLANs	17th Ka and Broadband Comm Conf, Palermo, Italy, 10/11	PDF
N. Ekiz A. Rahman P. Amer	Misbehaviors in TCP SACK generation	ACM Computer Communications Review, 41(2), 4/11 [Awarded 2011 Applied Networking Research Prize - http://irtf.org/anrp]	PDF
N. Ekiz P. Amer	A model for detecting transport layer data reneging	PFLDNeT 2010, Lancaster, PA, 11/10	PDF
N. Ekiz P. Amer P. Natarajan R. Stewart J. Iyengar	SCTP data acknowledgement with non-renegable selective acks (NR-SACKs)	IETF Internet Draft - draft-natarajan-tsvwg-sctp-nrsack	TXT
E. Yilmaz N. Ekiz P. Natarajan P. Amer J. Leighton F. Baker R. Stewart	Throughput analysis of non-renegable selective acknowledgments (NR-SACKs) for SCTP	Computer Communications, 33(16), 10/10, p1982-91	PDF
P. Natarajan F. Baker P. Amer J. Leighton	SCTP: What, Why, and How	IEEE Internet Computing, 13(5), 9/09	DOI PS
P. Natarajan	Leveraging transport services for improved application performance	PhD Dissertation, CIS Dept, Univ of Delaware, 2009 [Winner of UD 2009 Theodore Wolf Prize for the outstanding dissertation in Physical and Life Sciences.]	PDF
L. Budzisz R. Ferrus F. Casadeval P. Amer	On concurrent multipath transfer in SCTP-based handover scenarios	ICC 2009, Dresden, 6/09	PDF
P. Natarajan N. Ekiz P. Amer R. Stewart	Concurrent multipath transfer during path failure	Computer Communications, 32(15), 5/09, p1557-87	PDF
P. Natarajan N. Ekiz E. Yilmaz P. Amer J. Iyengar R. Stewart	Non-renegable selective acks (NR-SACKs) for SCTP	ICNP 2008, Orlando, 10/08	PDF
P. Natarajan P. Amer R. Stewart	Multistreamed web transport for high latency networks	2nd ACM SIGCOMM NSDR Workshop, 10/08	PDF
P. Natarajan N. Ekiz P. Amer J. Iyengar R. Stewart	Concurrent multipath transfer using SCTP multihoming: Introducing the potentially-failed destination state	Networking 2008, Singapore, 5/08	PDF
R. Bickhart P. Amer R. Stewart	TCP-to-SCTP translation shim layer in the FreeBSD kernel	EuroBSDCon2007, Copenhagen, 5/07	PDF
J. Iyengar P. Amer R. Stewart	Performance implications of a bounded receive buffer in concurrent multipath transfer	Computer Communications, 30(4), 2/07, p818-829	PDF
R. Stewart I. Rodriguez  K. Poon  A. Caro  M. Tuexen	Stream Control Transmission Protocol (SCTP) Specification Errata and Issues	IETF RFC4460, 4/06	TXT
P. Natarajan J. Iyengar P. Amer R. Stewart	Concurrent multipath transfer Using Transport layer multihoming: Performance during network failures	MILCOM 2006, Wash, DC, 10/06	PDF
J. Iyengar P. Amer R. Stewart	Concurrent multipath transfer using SCTP multihoming over independent end-to-end paths	IEEE/ACM Trans on Networking, 14(5), 10/06	PDF
J. Iyengar	Concurrent multipath transfer using SCTP multihoming	PhD Dissertation, CIS Dept, Univ of Delaware, 2006	PDF
P. Natarajan J. Iyengar P. Amer R. Stewart	SCTP: An innovative transport layer protocol for the web	WWW 2006, Edinburgh, Scotland, 5/06	PDF
A. Caro P. Amer R. Stewart	Retransmission policies for multihomed transport protocols	Computer Communications, 29(10), 6/06, p1798-1810	PDF
A. Caro P. Amer R. Stewart	Rethinking end-to-end failover transport layer multihoming	Annals of Telecommunications, 61(1-2), 1/06	PDF
J. Iyengar P. Amer R. Stewart	Receive buffer blocking in concurrent multipath transfer	IEEE Globecom 2005, St. Louis, 11/05	PDF
R. Bickhart	SCTP shim for legacy TCP applications	MS Thesis, CIS Dept, Univ of Delaware, 8/2005	PDF
A. Caro	End-to-end fault tolerance using transport layer multihoming	PhD Dissertation, CIS Dept, Univ of Delaware, 8/2005	PDF
P. Natarajan P. Amer R. Bickhart S. Ladha	Corrections on: Improving multiple file transfers using SCTP multistreaming	NOTE: This paper corrects invalid results originally publsihed in IPCCC 2004.  (PDA 6-6-05)	PDF
J. Iyengar P. Amer R. Stewart	Concurrent multipath transfer using SCTP multihoming	tech report	PDF
J. Iyengar P. Amer R. Stewart	Retransmission policies for concurrent multipath using SCTP multihoming	IEEE ICON 2004, Singapore, 11/04	PDF
A. Caro P. Amer R. Stewart	End-to-end failover thresholds for transport layer multihoming	MILCOM 2004, Monterey, 11/04	PDF
J. Iyengar P. Amer R. Stewart	Concurrent multipath transfer using transport layer multihoming: Performance under varying bandwidth proportions	MILCOM 2004, Monterey, 11/04	PDF
A. Caro P. Amer R. Stewart	Retransmission schemes for end-to-end failover with transport layer multihoming	GLOBECOM 2004, Dallas, 11/04	PDF
S. Ladha P. Amer A. Caro J. Iyengar	On the prevalence and evaluation of recent TCP enhancements	GLOBECOM 2004, Dallas, 11/04	PDF
R. Stewart P. Amer	SCTP	Internet Society Brief 17, 6/04	PDF
J. Iyengar K. Shah P. Amer R. Stewart	Concurrent multipath transfer using SCTP multihoming	SPECTS '04, San Jose, 7/04	PDF
S. Ladha S. Bauke R. Ludwig P. Amer	On Making SCTP Robust to Spurious Retransmissions	ACM Computer Communication Review, 34(2), 4/04	PDF
S. Ladha P. Amer	Improving multiple file transfers using SCTP multistreaming	NOTE: The original IPCCC results were invalid; corrected results are now available (PDA 6-6-05) Corrected results appear here IPCCC '04, Phoenix, AZ, 4/04
A. Caro J. Iyengar P. Amer G. Heinz S. Ladha K. Shah	SCTP: A Proposed Standard for Robust Internet Data Transport	IEEE Computer, 36(11), 11/03, 56-63	PDF
J. Iyengar K. Shah P. Amer R. Stewart	Concurrent Multipath Transfer Using SCTP Multihoming	Tech Report 2004-02, CISC Dept, Univ of Delaware, 2003	PDF
G. Heinz	Priorities in SCTP Multistreaming	MS Thesis, Tech Report 2004-01, CISC Dept, Univ of Delaware, 2003	PDF
A. Caro K. Shah J. Iyengar P. Amer R. Stewart	SCTP and TCP Variants: Congestion Control Under Multiple Losses	Tech Report 2003-04, CISC Dept, Univ of Delaware, 2003	PDF
S. Ladha P. Amer A. Caro J. Iyengar	Improving file transfer in FCS networks	MILCOM '03, Boston, MA, 10/03	PDF
A. Caro P. Amer R. Stewart	Transport layer multihoming for fault tolerance in FCS networks	MILCOM '03, Boston, MA, 10/03	PDF
A. Caro P. Amer R. Stewart	Retransmission policies with transport layer multihoming	ICON '03, Sydney, Australia, 9/03	PDF
J. Iyengar A. Caro P. Amer G. Heinz R. Stewart	Making SCTP more robust to changeover	SPECTS '03, Montreal, Canada, 7/03	PDF
A. Caro J. Iyengar P. Amer G. Heinz	Using SCTP Multihoming for Fault Tolerance and Load Balancing	SIGCOMM 2002 Student Poster - ACM Computer Communication Review, 32(3), 7/02	ABS PDF
J. Iyengar P. Amer  R. Stewart  I. Rodriguez	Preventing SCTP Congestion Window Overgrowth During Changeover	IETF Internet Draft (work in progress) draft-iyengar-sctp-cacc-01.txt	TXT
J. Iyengar  A. Caro  P. Amer  G. Heinz  R. Stewart	Making SCTP More Robust to Changeover	Tech Report 2003-01, CISC Dept, Univ of Delaware, 2002	PDF PS
S. Iren P. Amer	Application Level Framing Applied to Image Compression	Annals of Telecommunications,57(5), 5/02	PS
P. Conrad  G. Heinz  A. Caro  P. Amer J. Fiore	SCTP in Battlefield Networks	MILCOM '01, Washington, DC, 10/01	PDF
P. Conrad  A. Caro  P. Amer	ReMDoR: Remote Multimedia Document Retrieval Over Partial Order Transport	ACM Multimedia '01, Ottawa, Canada, 9/01	PDF
P. Conrad	Partial order/reliability transport for multimedia applications	PhD Dissertation, CIS Dept, Univ of Delaware, 2000	PDF-zip
S. Iren	Network-Conscious Image Compression	PhD Dissertation, CIS Dept, Univ. of Delaware, 1999	PS-gzip
A. Caro  P. Amer  P. Conrad  G. Heinz	Improving Multimedia Performance Over Lossy Networks via SCTP	ATIRP '01, College Park, MD, 3/01	PS
A. Caro  P. Amer  S. Iren  P. Conrad	Web-Integrating Network-Conscious Image Transmission	ATIRP '00, College Park, MD, 3/00	PDF PS
S. Iren P. Amer	SPIHT-NC: Network-conscious zerotree encoding	2000 Data Compression Conf., Snowbird, UT, 3/00	PDF PS
S. Iren  P. Amer  P. Conrad	The Transport Layer: Tutorial and Survey	ACM Computing Surveys, 31(4), 12/99	PDF-1 PDF-2 PS-1 PS-2
M. Steinder  S. Iren  P. Amer	Progressively Authenticated Image Transmission	MILCOM '99, Atlantic City, NJ, 11/99	PDF PS
P. Amer  S. Iren  G. Sezen  P. Conrad  M. Taube  A. Caro	Network-conscious GIF image transmission over Internet	Computer Networks, 31(7),  4/99	PDF PS
S. Iren  P. Amer  A. Caro  P. Conrad  G. Sezen  M. Taube	Network-conscious compressed image transmission over battlefield networks	MILCOM '98, Boston, MA, 10/98	PDF PS
P. Conrad  P. Amer  M. Taube  G. Sezen  S. Iren  A. Caro	Testing environment for innovative transport protocols	MILCOM '98, Boston, MA, 10/98	PDF PS
S. Iren  P. Amer  P. Conrad	NETCICATS: Network-conscious image compression and transmission system	Lecture Notes in Computer Science, 1508, Springer Verlag, 9/98	PDF PS
S. Iren  P. Amer  P. Conrad	Network-conscious compressed images over wireless networks	Lecture Notes in Computer Science, 1483,  Springer Verlag, 9/98	PDF PS
R. Marasli  P. Amer  P. Conrad	Metrics for quantifying partially ordered transport services	6th Int'l Conf. on Telecommunication Systems, Nashville, 3/98
E. Golden	TRUMP - Timed Reliable Unordered Message Protocol	CISC Dept MS Thesis, 12/97
R. Marasli  P. Amer  P. Conrad	Partially reliable transport service	IEEE ISCC '97 - 2nd Symp. on Computers and Communications, Alexandria, Egypt, 7/97	PDF PS
P. Conrad  P. Amer  E. Golden  S. Iren  R. Marasli  A. Caro	Transport QoS over unreliable networks: no guarantees, no free lunch!	5th Workshop on QoS, Columbia Univ., 5/97	PDF PS
R. Marasli  P. Amer  P. Conrad	An analytic study of partially ordered transport services	Computer Networks and ISDN Systems, 29(6), 5/97	PDF PS
P. Amer  P. Conrad  E. Golden  S. Iren  A. Caro	Partially ordered, partially reliable transport service for multimedia applications	ATIRP '97, College Park, MD, 1/97	PDF PS
R. Marasli  P. Amer  P. Conrad	Optimizing partially ordered transport services for multimedia applications	Multimedia Modeling: Towards The Information Superhighway, (Courtiat, Diaz, Senac, eds), World Scientific Pub Co, 1996	PDF PS
R. Marasli  P. Amer  P. Conrad	Retransmission-based partially reliable transport service: An analytic model	IEEE INFOCOM 96, San Francisco, 3/96, 621-629	PDF PS
P. Conrad E. Golden  P. Amer  R. Marasli	A multimedia document retrieval system using partially ordered, partially reliable transport service	Multimedia and Computing Networking '96, San Jose, CA, 1/96	PDF PS
P. Conrad  P. Amer  R. Marasli	Graceful degradation of multimedia documents via partial order and partial reliability transport protocols	IEEE Workshop on Multimedia Synchronization, Virginia, 5/95	PDF PS
T. Connolly  P. Amer  P. Conrad	An Extension to TCP: Partial Order Service.	IETF RFC 1693, 11/94	TXT
R. Marasli  P. Amer  P. Conrad  G. Burch	Partial order transport service: an analytic model	9th IEEE Workshop on Computer Communications, Marathon, FL, 10/94	PDF PS
P. Amer T. Connolly  C. Chassot  P. Conrad M. Diaz	Partial order transport service for multimedia and other applications	IEEE/ACM Trans on Networking, 2(5), 10/94	PDF PS
M. Diaz  A. Lozes  C. Chassot  P. Amer	Partial order connections: a new concept for high speed and multimedia services and protocols.	Annals of Telecommunications, 49(5-6), 5/94
P. Amer  C. Chassot  T. Connolly  M. Diaz	Partial order transport service for multimedia applications: Unreliable service	INET '93, 3rd International Networking Conf., San Francisco, 8/93
P. Amer  C. Chassot  T. Connolly M. Diaz	Partial order transport service for multimedia applications: Reliable service	2nd Conf. on High Performance Distributed Computing, Spokane, WA, 7/93